Start here
How to use this guide
The CBAT is a long day made up of short, unfamiliar tests. Nearly everything that goes wrong on it goes wrong for one of two reasons. Either you practised something that doesn’t match the real test, or you spent your attention defending a task that was always going to fall apart. I’ve built this guide around those two problems.
194
specific findings,
graded for confidence
graded for confidence
23
entries covering all
24 official tests
24 official tests
72
well established
by several candidates
by several candidates
102
resting on a single
account so far
account so far
How sure am I about each point?
Every point carries a grade instead of being presented as equally solid. That matters here, because some of what gets passed around about the CBAT is reliable, some rests entirely on one person’s memory, and a bit of it is just out of date.
✓ Well established! Single account✕ Outdated? From the published guides
- Well established. Backed up by several people who’d actually sat the test, usually at different times. Plan around these. 72 points.
- Single account. One source, or people talking about practice instead of the real thing. Often the most useful detail here, and also the most likely to be wrong. Check these during your reading and practice time. 102 points.
- Outdated. Still repeated everywhere, but it doesn’t hold any more. Don’t calibrate against these. 1 points.
- From the published guides. Taken from the published test guides rather than from anyone who sat it. These are here because six of the twenty-four tests have almost no first-hand information behind them, and a described test beats a blank page, but nobody has confirmed them. 19 points.
What this is, and what it isn’t. The tests get revised, details vary
between sittings, and memory is imperfect. People write these accounts up hours after a long, stressful
day. So this is a preparation aid built from a lot of overlapping accounts. It isn’t official, and it
doesn’t reproduce live test material. Where accounts contradict each other I’ve said so instead of quietly
picking a winner.
The short version
Twelve things worth knowing before anything else
- The pass mark for pilot is 112 out of 180, not 128. And "you need 130 to be competitive" is a leftover from when there were more applicants than places. A pass is what’s required.
- Every domain carries its own minimum, and one stanine short fails you for that role. Even with a total well clear of the mark. This catches more people than any other single thing, and it’s why you can’t write off a test you’re bad at on the basis that the rest will carry it.
- The real test is usually easier than practice on SkyWatch and similar tools. It runs slower, the numbers are rounder, and you get a proper allowance per question. Drowning in practice predicts nothing.
- Read every instruction fully, twice, and use the whole practice run. Nobody scores you for getting through the briefings quickly, and this is the advice people who passed repeat most.
- On the bearings test, the number you type is the heading you want the aircraft to end up flying. Not the size of the turn, and not a mark on the ring printed round the display. North is the same everywhere on the screen, so 090 always means “settle on an easterly heading” whichever way the aircraft happens to be pointing at the time, and the L/R you pick only decides which way round it rotates to get there. Forums write this as each aircraft having “its own north”, which just means a north line drawn through the aircraft parallel to the display’s north. The frame does not turn with the aircraft.
- In FLAG, only the circled aircraft matter. The uncircled callsigns are there purely to pull you off.
- You’ll lose the callsigns in FLAG near the end, so let it happen. Bank the maths and the shapes instead. People who did that still scored well.
- No pen, no paper, for anything. Every calculation is mental, which is why estimating and rounding well beats being precise.
- Feeling like you’ve failed a test means nothing. Nearly everyone who described walking out of one convinced they’d ruined it went on to pass, including a bloke who nearly left at lunchtime.
- Drill mental arithmetic away from the apps. It’s the one thing app practice won’t do for you, and it feeds into several different tests.
- Sleep counts, even though nobody marks it. Hot rooms, springy mattresses and nerves are behind at least one failure I know of.
- Use every break, and never replay the test you’ve just done. That’s how one bad exercise turns into a bad day.
The tests
Test by test
All twenty-four official tests are here, in 23 entries, because Trace 1 and Trace 2 share one. Each starts with a What the test is panel describing the format itself, taken from the published CBAT TMI and Air Defence Academy test guides, then my overall read, then the specific points. The panel tells you what you’re walking into; everything under it is what people who sat it actually found. They’re kept separate on purpose. Tests are ordered by how much reliable information exists, so the best documented come first and the blind spots come last.
What the test isRoughly ten minutes of multitasking with a short break at the halfway point, running three tasks at once. Figures: quickfire mental arithmetic answered on an on-screen keypad in about four seconds a question. Logistics: watch the circled aircraft and press the matching key as they cross green circles, yellow triangles or red rectangles. Groups: a two-letter callsign flashes up with a beep as aircraft enter and leave, and you get yes/no questions on which are still on screen. The load climbs throughout.
Format description from the published CBAT TMI test guides
(rafcbat.wordpress.com) and the Air Defence Academy test guides
(airdefenceacademy.com/testguides).
Probably the highest-stakes test on the CBAT, and the one that falls apart fastest once it gets going. If you’ve only got time to drill one thing properly, make it this.
You only have to recall the callsigns of aircraft with a circle around them. The uncircled ones carry deliberately similar callsigns to pull you off, so ignore them completely.
This is the most important fix to make to your practice, because some practice apps have you recalling both. They’ll tell you clearly on the day, but if you already know it you won’t be relearning the task while the clock runs.
The aircraft move fast and erratically, and it gets busier as it goes.
Everyone says this. Your timing will feel wrong for the first thirty seconds or so. Expect that and it won’t throw you.
Almost everyone loses the callsigns by the final round. The play that works is to bank the maths and the shapes, then answer callsign questions on gut familiarity.
This is the most consistent advice from people who passed. Several gave up tracking callsigns entirely near the end, guessed on whether one "felt familiar", kept on top of everything else, and still scored well. Don’t burn attention defending the bit that was always going to collapse.
Take your points early. The first 80% is manageable, so go hard at all three task types while you’ve still got spare capacity.
The difficulty ramps up, it doesn’t start high. Points you pick up in the calm phase cost you far less than points you fight for at the end.
The on-screen instructions say you get partial points for “near misses” on shape crossings.
I’ve only seen this reported once. If it’s right, a marginal press beats skipping. Don’t build your whole approach on it though.
FLAG is the only test feeding the central information processing domain, so nothing else on the day can cover for it.
Two candidates confirmed this from their own results pages. Because it stands alone in its domain, a weak FLAG fails you for the role outright no matter how good your total is, and that is exactly what happened to at least one candidate, who missed narrowly and lost all three aircrew roles he wanted. If you only drill one test properly, this is the one.
The shapes are reasonably spread out but not big, so at that aircraft speed you sometimes get almost no window to press. Worst where a track clips the tip of a triangle.
So watch the tips. That’s where presses get missed.
At some point late on it can throw a huge number of aircraft up at once. One candidate counted something like thirty on screen.
He also went to the boarding officer expecting a sheet full of red and found he had done far better than he feared. So treat the spike as the test doing its job rather than as evidence you’ve lost it.
Don’t assume FLAG will show up if you’re sitting a different service’s version. At least one overseas candidate didn’t get it at all.
Their version seems to swap in a different test. There’s more on this further down.
There are two different sounds, one for a circled aircraft joining the screen and one for it leaving.
Learn which is which during the practice run. Treat the sound as a cue to glance at the screen, not something to keep count of.
What the test isSpeed, distance and time problems built around a route map. A left-hand menu holds the data charts, the map shows the route between named checkpoints, and a solving table sits underneath. You read speed off a bar chart against package weight, take distances from the map, and work time as distance ÷ speed, about one minute per question including reading time. Bad weather between checkpoints adjusts your speed by a percentage rather than the distance, and partial credit is given for close answers, so rounding pays.
Format description from the published CBAT TMI test guides
(rafcbat.wordpress.com) and the Air Defence Academy test guides
(airdefenceacademy.com/testguides).
The best documented test there is. The arithmetic is friendlier than people expect. The comprehension is the hard part.
The numbers in the real ANT are round and friendly.
Everyone who’s sat the test agrees on this. If your practice sums feel tougher, you’ll turn up over-trained on the arithmetic, which is exactly where you want to be.
The real questions carry far more variety and more steps though: storms and weather, two aircraft at once, fuel calculations, and traps in the wording.
A classic trap is a question phrased as "Alpha to Bravo is 67 miles" when the table says 43 miles remaining. The maths isn’t what fails people here. Misreading the question is. Read the last line of the objective box first, because that tells you what’s actually being asked.
Two aircraft turn up later in the test, usually called Pluto and Jupiter, with different speeds and a separate chart each.
Typical question: one leaves point A with no package at 15:30, so what time does the other have to leave point B carrying 500kg for both to reach point C together? Check you’re reading the right aircraft’s chart. Careless errors cluster here.
You get a full minute per question, plus reading time at the start.
So you’re not in a race here. Use the minute.
Speed is in miles per minute and depends on package weight. Fuel consumption is given per distance and depends on speed.
So a fuel question is one extra lookup, not one extra calculation. Weight gives you speed, speed gives you consumption, then you work it exactly like a time question. Knowing that beforehand takes most of the sting out of it.
Get to comfortable full marks on speed, distance and time practice, on SkyWatch or a similar tool, before test day.
Two people who passed landed on the same target independently. The score isn’t really the point. Once the arithmetic is automatic you’ve got spare capacity for the wording, and the wording is what catches people.
You get three tabs: a short intro telling you to round and estimate, a speed against load chart, and a speed against fuel consumption chart. Plain bar charts, whole numbers.
One detailed first-hand description, and it fits everything else I’ve seen. Only one person described the tab layout though, so hold the specifics loosely.
Speed and fuel consumption values look to be single-digit whole numbers, and no single question asks for both a time and a fuel figure.
Both from one account, and hedged when it was given. Handy if you’re writing your own practice questions. Not something to lean on.
One question gives you a tailwind and makes you faster. The other wind questions slow you down. Your speed is always stated, and if it isn’t, assume the maximum for the parcel weight.
Read any weather question twice, because the effect flips direction.
Bad weather is flagged by a thunderstorm icon on the leg, and the reduced speed is given to you in the table rather than left for you to work out.
So a storm question is a lookup, not a calculation, so go and find the new figure instead of trying to apply a penalty yourself. One candidate wasn’t certain the slower speed was stated every single time, so if you genuinely can’t find it, fall back on the stated maximum for the weight and move on rather than burning the minute hunting.
Some questions ask for half the journey rather than all of it: time to the halfway point, or from it.
Easy marks and an easy trap in equal measure. The method doesn’t change at all, you just halve the distance before you start. It catches people because it’s a single word in the wording, which is exactly the kind of thing you skim past when you’re working at pace.
The weight column steps in round hundreds (100kg, 200kg, 300kg and so on) but the weight in the question can be an untidy number like 220kg.
You are meant to round to the nearest row and read the speed off it. There is no interpolating and no exact answer waiting to be found, which is worth knowing in advance because the instinct on seeing 220 is to assume you’ve misread the chart.
Press enter to commit your answer. The minute runs out whether or not you’ve submitted, and an answer you worked out but never entered scores nothing.
One candidate lost a question exactly this way. It sounds trivial and it is, right up until you spend fifty seconds getting a hard one right and then don’t bank it. Get into the habit of entering a rough figure early and correcting it, rather than typing only once you’re sure.
The route maps get bigger as the test goes on, with more checkpoints and more possible routes, but the route you’re supposed to take is always spelled out.
So a busier map isn’t a harder navigation problem, it just looks like one. Read off the route you’ve been given and ignore the rest of the picture.
You can backspace while typing an answer in the real test. Some practice apps won’t let you.
Small thing, but it removes a bit of panic that your practice may have trained into you.
Rounding is safe. Answers a couple of minutes off the exact figure still mark as correct.
Widely repeated, but I’ve never seen anyone quote the official tolerance. Round with confidence, just don’t get sloppy on the assumption that the margin is huge.
Some of the speed-distance-time practice tools floating around online are far harder than the real test.
Fine for overtraining, useless for judging where you are. If one of them is wrecking your confidence, that’s the tool, not you.
The practice tools disagree about which way round the speed chart runs. Some show a heavier package flying slower, at least one published chart shows the opposite.
The first-hand accounts point at heavier being slower. One describes 7 miles a minute as the speed carrying a 200kg package, dropping as the load goes up. It matters less than it looks, because you read the number off the chart in front of you on the day rather than recalling it. What you should not do is drill until the relationship feels automatic, then meet a chart that runs the other way.
The real test gives partial credit for close answers, so being a few minutes out is not a zero.
Some practice apps only accept exact answers, and a run of zeroes on one can rattle you when your method is basically sound. Judge yourself on whether you got there inside the minute with a sensible estimate, not on whether an app lit up green.
Expect wordy questions with several factors at once: weather, fuel running low, and the two-aircraft stage at the end.
The numbers themselves stay friendly. It’s the reading that grows, which is one more reason to get the arithmetic automatic before the day.
ANT is the test most often named by candidates who did well everywhere else and still missed a role by a single domain.
If your arithmetic is shaky, this is the best place to spend extra practice time, because the rest of the day can’t make up for it.
What the test isForty rapid-fire arithmetic questions, one operator at a time. There is no backspace and no second chance once a digit is entered. Nothing to interpret: it is pure speed and accuracy against the clock.
Format description from the published CBAT TMI test guides
(rafcbat.wordpress.com) and the Air Defence Academy test guides
(airdefenceacademy.com/testguides).
Short, brutal, and the one test you can properly train for away from any app.
Drill mental arithmetic away from the apps. Times tables, addition, subtraction, multiplication.
Everyone who described their prep in any detail did this separately from app practice, including the candidate who failed twice and then passed everything on his third go. He was blunt that the app on its own hadn’t been enough.
Plenty of app practice still might not make you fast enough. One high scorer ran out of time on questions after spending about twenty seconds on 138 ÷ 4.
Awkward division is the weak spot. Practise the ugly sums, not the tidy ones.
It runs for roughly a minute, and it’s realistically the only test where you’ll type with both hands.
Everywhere else, one hand is usually busy with something.
Round to a friendly multiple, then fix the remainder.
Say 321 ÷ 7. Take 350 ÷ 7 = 50, then knock off 28 ÷ 7 = 4, so about 46. Or 183 ÷ 4: 160 ÷ 4 = 40, plus 20 ÷ 4 = 5, so 45. This one comes up again and again, and people find it clicks straight away.
What the test isSix aircraft displays (Message, Engine, Navigation, Sensor, Mission and System) of which only two can be open at once, switched from a row of index buttons. A warning panel and a clock stay fixed at the top, and the primary objective is to keep that warning panel empty: hold three fuel tanks within 50 litres of each other, keep airspeed within ±10 kt of the required figure, run the air and ground sensors on their 2- and 4-minute schedules, complete load-drop and video entries, and keep hydraulic pressure between 90 and 110 while entering comms codes before their countdowns expire. Scheduled events need you on the right display within five to ten seconds of the timestamp.
Format description from the published CBAT TMI test guides
(rafcbat.wordpress.com) and the Air Defence Academy test guides
(airdefenceacademy.com/testguides).
The best news in this whole guide. Everyone who sat it found the real version noticeably easier than practice on SkyWatch and similar tools.
The real test runs slower and feels lighter than practice on SkyWatch and similar tools. More than one person had genuine idle moments on the day after feeling like they were drowning in practice.
Some of that is pacing and some of it is just ergonomics. A mouse on a desktop screen beats fingers covering half a phone. If practice is overwhelming you, that says nothing about how test day will go.
Keep one of your two windows permanently on the warnings and message panel, and use the other for tasks. Never close the warnings view.
The most repeated CUT tip there is, and near enough unanimous among people who passed.
You’re marked on speed, accuracy and warning count together. As few warnings as you can manage, cleared fast, correct inputs at the right time, airspeed held in range.
You get penalised for how long you sit in a warning state, not just for triggering one. A quick fix costs you very little. A slow one costs you a lot.
Pressure takes the longest to clear, so it needs the closest watch. Airspeed is the one that drifts out of limits easiest.
Reported often, though mostly by people talking about practice. The one person who mentioned his real-test warnings said pressure was the only thing he got pulled up on.
Type the codes on the objective page as fast as you can, and aim for two or three correct codes per line.
A real target beats a vague "do the codes", and it makes it much easier to tell whether your practice is going anywhere.
One task has you copy a position and a time out of separate messages into boxes, then release at exactly the time given.
Recent candidates said the number of displays matched SkyWatch and most of the other tasks felt easier. Watch the clock for that one release.
What the test isFour target types monitored simultaneously. Light: a box shows three coloured lights and a panel cycles colours at random; you press when they match. Scan: a named vehicle (tank, aircraft, two-wheeler) must be picked out of a panel that a scan line reveals only in slivers. System: an alphanumeric code has to be found in a scrolling list, 19 rows visible at a time in the real test. Scene: click every instance of a specified target in the central map panel, identified by colour, shape and direction of travel, ignoring the distractor shapes.
Format description from the published CBAT TMI test guides
(rafcbat.wordpress.com) and the Air Defence Academy test guides
(airdefenceacademy.com/testguides).
The test with the most going on at once, so learn every target type before the day.
The real test adds an “unknown” unit, shown as a yellow diamond, plus flashing red priority targets that appear for a few seconds and have to be clicked straight away.
Easy to be caught out by if you haven’t met them before. Go in knowing they exist so they don’t cost you the whole first round.
Instructions get specific. You might get "damaged friendly truck heading north", and then only those count.
It isn’t always "click all enemy tanks". Read each instruction properly. Over-clicking hurts you as much as missing.
It runs as eight or nine rounds that build, each one adding more objects to find and more items in the list panel.
It starts gently, so use the early rounds to lock in your scan pattern before the load climbs.
Anchor on the list panel codes, because they’re the only thing running constantly. Keep colours and vehicles in your peripheral vision, and clear map tasks the second they appear.
The logic is that lights and vehicles only match now and then, and map tasks are instant, while the codes never stop. Several people worked out this same order on their own, which is usually a sign it’s right.
The Scan panel shows only a thin slice of the vehicle at a time, sliding past as if through a letterbox, and the slice is enlarged and a little blurry.
Several people who sat it described it this way. Some vehicles look almost the same until about half the shape has gone past, so it takes more time away from the other panels than you expect. Don’t commit on the first sliver unless the shape is unmistakable.
What the test isTwo parts. Part 1 shows an attitude indicator and a compass at the top and four aircraft images below; you pick the image that matches both readings at once. Part 2 shows six instruments (altimeter, attitude, speed, ascend/descend, heading and turn) with five written statements, and you pick the statement that fits. The published method is to start from the statements, find the data point that separates them, and read only the dial that settles it.
Format description from the published CBAT TMI test guides
(rafcbat.wordpress.com) and the Air Defence Academy test guides
(airdefenceacademy.com/testguides).
The most trainable test on the CBAT. Everyone agrees on the method, so the only thing that varies is how fast you can run it.
Work by elimination from the answers instead of reading the gauges. Take each option in turn and bin it the moment one stat doesn’t match.
Don’t read the instruments first and then go hunting for an answer that fits. Start at the options, find the stat that differs between them, and check only that. People report jumping to near full marks the same day they switch over.
The multiple choice format shifts slightly on every question in the real test. Some practice apps keep one fixed layout.
This matters more than it sounds. A fixed layout quietly trains you to look at one spot on the screen, and when that moves every question you lose seconds before you even notice why.
Reportedly about five minutes for twenty questions, so roughly ten seconds each. That’s reachable after a few weeks of daily drilling.
Both accounts of the timing were second-hand, so treat ten seconds as a training target and not a confirmed limit. It’s realistic though. People have got there on about ten minutes a day for three weeks.
The classic mistake is ascending versus descending. Every other stat matches, so elimination won’t save you if you skim that word.
Give the climb wording a deliberate half-second check before you commit.
Some of the harder practice sites make all the options deliberately similar and give you ten seconds, which makes elimination too slow to work.
On those, read the gauges first and then scan for the match. Different tool, different best method, so don’t assume your technique carries over.
What the test isA 2D overhead radar picture with three aircraft: CA-A (red) and CA-N (yellow) flying ordered gate sequences (lettered waypoints for one, numbered for the other) plus a green interceptor chasing a moving enemy. All three continue on whatever heading you last assigned. Aircraft must keep at least 3,000 ft of vertical separation from each other and 1,000 ft clear of danger-area altitudes, shown as red circles: white outline for 2,000 ft, black outline for 3,000 ft. Interception works off the projected crossing point, not the enemy’s current position.
Format description from the published CBAT TMI test guides
(rafcbat.wordpress.com) and the Air Defence Academy test guides
(airdefenceacademy.com/testguides).
Comfortably the most feared test on the CBAT, the worst explained on the day, and the one people most often pass while being certain they’ve failed it. Read this section twice.
The number you type is the compass heading you want the aircraft to END UP flying. It is an absolute bearing, measured against the same fixed north for every aircraft on the screen. It is not the size of the turn, and it is not a point on the edge ring.
This is the most useful single thing on this page, so it is worth being blunt about the two ways people get it wrong. First, it is NOT relative to the heading the aircraft is already on: typing 090 does not swing it 90° from wherever it happens to be pointing, it puts it on an easterly heading. Second, it is NOT an instruction to fly at the 090 mark painted round the edge: a heading is a direction and that mark is a place, so the two only line up when the aircraft is sitting on the centre point. The L/R you pick is the only part of the input that is relative to anything, and it chooses which way round the aircraft rotates to reach the heading you asked for, not how far it goes. You will see this written on forums as each aircraft having “its own north”. That means the same thing, a north line drawn through the aircraft parallel to the display’s north, but the phrase misleads people into thinking the frame turns with the aircraft. It does not. North is north everywhere on the display. The numbers round the edge are a scale reference, handy for judging how big a 100° turn looks, and otherwise a distraction.
Flying west on 270, you select L and type 090. The aircraft turns left through south and settles heading due east on 090, a full 180° turn. It does not end up heading south.
This is the case that catches everybody, because 090 looks like it ought to mean “90 degrees”. Nothing about the input is measured against the heading you are already on, so a big number can produce a small turn and a small number a big one. If what you actually wanted was 90° of left turn from 270, the heading that lands on is 180, so 180 is what you type. Work out the heading you want first, then type that. Never type the size of the turn.
Everyone rates this the hardest and worst explained test on the CBAT. The instructions don’t make the controls clear.
Knowing that beforehand is worth something on its own. People who walked in expecting confusion handled it far better than the ones who assumed being confused meant they’d missed something obvious.
You can score well on DPT, and pass overall, while feeling like you completely botched it. The markers clearly don’t expect anyone to nail this one.
One candidate got reset to the gate again and again, missed an intercept entirely, felt awful about the whole thing, and still scored well on that exact test. Another seriously thought about walking out at the halfway point and passed for both roles he wanted. Whatever happens here, finish the day.
It builds. You start directing one aircraft through lettered gates, then juggle up to three at once, then there’s an intercept where you bring a marker onto a jet that’s manoeuvring.
Aircraft also have to keep vertical separation from each other and stay clear of marked danger areas. The early phase is where you can afford to get things wrong.
Hitting “clear” to fix a mistyped input also wipes the direction you’d selected. Re-enter it, or the aircraft will swing round the long way.
One person reported this and said it threw him more than anything else on the test. Costs you nothing to check during your practice run and could save the whole exercise.
The danger areas are fixed. They don’t drift and they don’t disappear, so you can take them off your plate in the opening seconds by putting both aircraft at an altitude that clears them and leaving them there.
One of the most useful things anyone has said about this test. It converts a running hazard into a one-off decision, and everything you save goes into the gates and the intercept. The one thing it doesn’t cover is the AI traffic crossing the scene, which can be sitting at whatever altitude you just picked, so you still watch for other aircraft, you just stop watching the ground.
Turning an aircraft the long way round is sometimes the right call. The extra seconds in the turn are seconds you can spend reading the rest of the screen.
Worth knowing because the instinct is always to take the short turn. When you’re behind on the picture, the slow turn is free thinking time rather than a mistake.
The practice attempt matters more here than anywhere else. Use it to answer one specific question, namely what exactly your input does to the aircraft, rather than treating it as a warm-up.
This is the test where walking in with an untested assumption about the controls costs the most, and the practice run is the only place you can settle it for free.
What the test isA passage of text and four numbered answers, with no diagram to work on. Two question types: static relationships, where you chain fixed positions between landmarks, sum the offsets along one axis and read the direction off the sign of the net value; and journeys, where you follow a sequence of turns and distances and give the direction or distance back to the start. Everything is mental: the day allows no pen or paper.
Format description from the published CBAT TMI test guides
(rafcbat.wordpress.com) and the Air Defence Academy test guides
(airdefenceacademy.com/testguides).
The test that quietly rewards school-level geometry, inside an assessment that’s officially supposed to be impossible to prepare for.
For "how far from the start" questions, split each leg into north/south and east/west components, then put them back together with Pythagoras.
Add up all the northward distance and take off the southward. Same again for east and west. Your straight line distance is then the square root of the two squares added together. There’s no shortcut beyond estimating, and people who tried to picture it with their fingers did noticeably worse than the ones who just did the sums.
It’s multiple choice, so round hard. Snap to the nearest square number instead of working out a precise root.
If your components give you something like 340,000, spotting that it sits closer to 600² than 500² is all the precision you need. Knowing your squares up to about 900 without thinking makes this test far easier.
Map the chain backwards, from the last point named.
Placing "D is 20km north of E" is much easier once you already know where E is. It takes most of the confusion out of the chains, and one bloke said it improved his prep straight away.
Questions use plain directions and distances. No compass bearings.
So you shouldn’t get anything like "200 miles at 100°". One person who sat it said this, and nobody has contradicted it.
Don’t use a chatbot to check your DAD answers.
People who tried got different answers on re-asking, and it agreed with whatever they claimed was right. Work them by hand, or check against a worked example you already trust.
What the test isA grid with columns 0–9 and rows A–J, each cell representing 2 km. Units appear as coloured icons with labels: yellow friendly, red hostile, white unknown. A right-hand panel carries the controller aircraft’s next waypoint, next waypoint time, altitude and comms channel. Information arrives through three unsynchronised channels at once (spoken audio, on-screen text and imagery) and then everything disappears and you answer multiple choice from memory. Grid references are given chess-style, column first then row.
Format description from the published CBAT TMI test guides
(rafcbat.wordpress.com) and the Air Defence Academy test guides
(airdefenceacademy.com/testguides).
Widely agreed to be one of the hardest, and one of the few where people report getting partial credit.
Very hard, and there’s not much you can do to prepare beyond general memory technique.
People keep describing it as one of the toughest on the day. They also describe it as good fun, and plenty who found it brutal still passed.
You get credit for being close on the grid. Positions don’t have to be exact.
This drives the whole strategy below, and it’s why you shouldn’t spend your memory on precise coordinates.
Spend your memory on callsigns, unit numbers and altitudes. Treat grid position as roughly right.
Knowing broadly where something sat is good enough to score. A wrong callsign is just wrong. Some people also track units physically with a finger on the screen.
Questions go well beyond unit positions. Next waypoint times, altitude, comm level and channel, and direction are all fair game.
The range is deliberately wide, so don’t sink everything into one type of detail. If a value is on screen, assume you can be asked about it.
Not every question is multiple choice. One candidate said close to half asked you to type the answer, such as a waypoint time or an altitude.
So aim to hold the exact figure, not a rough sense of it. Where questions were multiple choice, there could be more than four options.
Information can arrive one item at a time, a few seconds each, rather than as one full panel to study.
Accounts differ on how heavy it feels. Train on the harder setting and either version will feel manageable.
SAT can decide a role on its own. One candidate cleared the overall mark and still missed a role on this domain alone.
So it isn’t a test to write off as unpreparable. Memory technique and plenty of exposure to the format both help.
What the test isEight or nine tabs, each showing one isolated layer of the same landscape: farm position, truck, troops, trees, aircraft flight paths with their speed and altitude charts, helicopter paths with theirs. No tab shows the full picture, and the whole sequence runs for one minute, so the time per tab is not equal. Afterwards you get a 3D rendered video of the scene and about 50 seconds of true/false questions on it, with no replay.
Format description from the published CBAT TMI test guides
(rafcbat.wordpress.com) and the Air Defence Academy test guides
(airdefenceacademy.com/testguides).
Barely documented anywhere, but there’s one technique that makes a huge difference.
You study displays showing where things sit on a grid, then get a two or three second clip of an environment and have to say whether one specific detail in it was correct.
Something like "are the farms in this image correct?" As it goes on it adds more for you to learn, including aircraft, helicopters and their flight paths.
Only the detail being asked about has to be right. Everything else in the image can be wrong and it doesn’t matter.
This could be the difference between passing and failing, and it comes from one account, so confirm it during your reading time. Use fixed terrain like hills as reference points, find the thing they’ve asked about, and ignore the rest of the frame.
Some tabs show the map rotated, with the compass rotated to match, so you have to turn it in your head to line it up with the other layers.
Reported by more than one candidate. Check the compass on every tab before you study it. A layer learned the wrong way round is worse than one not learned at all.
A tab fades out if you stay on it too long, and there is a short fade every time you switch.
That makes flicking back and forth slow, so plan one steady pass through the tabs instead of hopping between them.
Later scenes add jets and a helicopter. Their tabs show the route and how speed and height change along it, and the questions can ask about any of the three.
Several accounts agree it isn’t just the route that gets tested. Learn the shape of each change: climbing or descending, speeding up or slowing down.
At least one candidate missed a role by a fraction in this domain after skimming the SIT instructions.
It is one of the least familiar formats on the day, so use every second of the briefing and the practice.
Some candidates say the scene is shown once. Others say it was shown twice, with the question already on screen.
Either way, read the question first if you get the chance and look for that one detail.
What the test isA numbered index of 15 tabs down the right and a dual-pane display area on the left, so two tabs are readable at once. Opening a third closes the oldest automatically. Each tab holds one category of data about a system, and roughly 30 multiple-choice questions require you to locate values across several tabs and calculate from them. No single tab answers a question. The whole thing runs in 20 to 25 minutes.
Format description from the published CBAT TMI test guides
(rafcbat.wordpress.com) and the Air Defence Academy test guides
(airdefenceacademy.com/testguides).
Nothing to practise on, but there’s one approach to the reading time that people swear by.
You get tabs describing a system, mixing figures and text, things like how much oil something burns or how much power it puts out. Then questions that make you find and use the right figure.
It’s a search and apply task, not a memory task. Nobody expects you to have learned the content.
Use the reading time twice. Once properly for the content, then a quick second skim just to fix which tab holds which kind of information.
Under time pressure the bottleneck is finding the right tab, not doing the arithmetic. Knowing where everything lives beats knowing the numbers.
What the test isA white ball inside a first-person tunnel, flown on a joystick and kept centred as the tunnel curves, rises and narrows. Wall contact is penalised in proportion to how long you touch. Coloured shapes appear in the tunnel to be flown through, or avoided when an audio rule says so. Stacked on top: pull the trigger on irregular beeps, change the ball’s displayed number on spoken command with your other hand on the keyboard, hold a spoken 4–6 digit sequence for later recall, and act only on instructions carrying your NATO phonetic callsign. Instructions play once, under deliberate background chatter, and tasks accumulate across rounds without resetting.
Format description from the published CBAT TMI test guides
(rafcbat.wordpress.com) and the Air Defence Academy test guides
(airdefenceacademy.com/testguides).
People describe it as good fun at the start and punishing by the end.
You fly a ball through hoops and pull the trigger when told, while the audio throws junk chatter at you along with real instructions for your callsign and fake ones using other callsigns.
The junk is overlapping voices reading filler. The whole discipline is filtering for your own callsign and acting only on that. Anything addressed to another callsign is a trap.
You also have to hold a six digit number, and later on the instructions come with a delay attached.
Twelve seconds in you might be told to change colour to red at fifty-eight seconds. So you’re holding a pending action while still flying and still filtering. Two separate candidates described this independently, which makes it one of the more solid things on this page.
The delay is given as a time on the clock, not as a countdown. You get something like "change your number to 9 at 23:00:30" and it is on you to watch for it.
This matters for how you hold it. A countdown you can feel; a timestamp you have to keep checking against the clock in the corner while doing everything else. Get into the habit of glancing at the clock on a rhythm rather than waiting for the moment to feel close.
Four things get asked of you: fly through or avoid the shapes, hold a code for recall, change your ball’s number and colour, and pull the trigger.
Useful as a checklist because it bounds the problem: nothing else is coming. One candidate reported his instructions being almost entirely about which shapes to avoid, with nothing about colour at all, so the mix you get may be narrower than the full list.
The trigger cue is a loud beep, not a spoken order.
Worth separating in your head from everything else, because it’s the one task that doesn’t come through the voice channel. You can be filtering chatter for your callsign and still have to react to a sound that has nothing to do with any of it.
The background noise is described as sounding like overlapping voices reading filler, or television adverts playing underneath.
It is deliberately the kind of audio your ear wants to follow. Knowing it is meaningless before you sit down makes it much easier to let it wash past instead of half-listening to it in case it matters.
It starts very easy and gets very hard.
Don’t read the opening stages as a sign of how it ends, and don’t let the late difficulty convince you you’ve fallen apart. The ramp is the design.
With a two-word callsign, recent candidates say you respond to either word on its own, not only to the pair said together, and the number of callsigns to listen for grows across the rounds.
Two separate accounts agree on this. Check it against the instructions on the day, because it is exactly the kind of detail that changes.
You type the code in while you are still flying, not at the end of the round.
Several candidates mentioned it. Practise keying with one hand while the other keeps the ball steady.
What the test isTrace 1: an animated aircraft in flight whose control inputs you mirror on the arrow keys as if you were flying it. You always control whichever aircraft is currently red, and two more (blue and yellow) join later; when the aircraft faces you or flies inverted, the pilot’s left is your right. Trace 2: up to five coloured aircraft manoeuvre in an animation and you then answer multiple choice on what happened: which entered or left the scene and in what order, which made the most turns, which never climbed.
Format description from the published CBAT TMI test guides
(rafcbat.wordpress.com) and the Air Defence Academy test guides
(airdefenceacademy.com/testguides).
Two completely different tests with confusingly similar names, and the difference isn’t what most candidates assume.
Trace 1 is copying the pilot’s control actions. Trace 2 is watching the scene and then answering questions on what you saw. It isn’t a second flying test.
Loads of candidates assume Trace 2 means estimating turns while you fly. It doesn’t, and going in with that idea puts your attention in completely the wrong place. Watch it properly and expect questions afterwards.
Trace 1 is the easy one. Trace 2 is the overwhelming one.
Reported in that order every time, and Trace 2 keeps showing up on candidates’ lists of worries alongside DPT and FLAG.
Trace 1 on SkyWatch and similar tools is said to be very close to the real thing.
So your practice here is a fair guide to how the real one will feel.
Plenty of people passed without practising either trace test. If you do want to drill Trace 2, the practice version on this site is the one to use.
Nothing else covered it, which is why I built one. It is my own version rather than the real subtest. What it trains is the part that catches people out: watching the scene actively because you know questions are coming, instead of spectating and then trying to reconstruct it afterwards.
Serious flight sim time is said to help with the 3D visualisation and the instrument work.
Note it gets reported as helping the spatial and instrument tests specifically, not the memory or arithmetic ones. It won’t replace targeted practice.
The trace tests are where tiredness hurts you most.
One candidate on a two-day overseas version put failing the trace and motor tests down to a sleepless night beforehand. Sleep counts for more on these than on the paper-style tests.
In Trace 1 the aircraft you follow is whichever one is red at the time, so control switches between aircraft, and it gets fast towards the end.
Watch for the colour change rather than sticking with one aircraft.
In Trace 2 aircraft can leave the scene and come back, and the questions range widely: who arrived and left, in what order, who turned most and who never climbed.
Keep a rough running order of who arrived and who left. It’s the question type people find hardest to reconstruct afterwards.
What the test isA red dot drifting continuously across the display and a crosshair fixed at its centre; the job is to keep the two aligned. The joystick handles the vertical axis: push away to bring the dot down, pull back to bring it up. Foot pedals handle the lateral axis, worked with forward ankle pressure like rudder pedals rather than pressing down: right pedal when the dot drifts left. The dot never holds position between corrections, so it is already moving while you respond to the last drift.
Format description from the published CBAT TMI test guides
(rafcbat.wordpress.com) and the Air Defence Academy test guides
(airdefenceacademy.com/testguides).
The only test where the kit itself is a known variable, and where you’re allowed to do something about it.
The stations aren’t identical. Sticks and pedals vary in dead zone, sensitivity and wear, and people regularly describe the pedals as sticky with low sensitivity.
It comes down to the settings profile loaded on each station and how much use it’s had. You’re not imagining it, and you’re not supposed to just put up with it.
If your station feels unreasonable or unusable, say so at the start. Staff keep spares. Spend a minute getting the pedals positioned properly before you begin, too.
Reported by someone who did exactly that. It’s a normal request, not a complaint, and a minute on setup is a lot cheaper than a whole exercise fought against bad ergonomics.
Take your shoes off before you use the pedals.
Straightforward advice from someone who passed, and it fits everything else people say about the pedals being stiff and unresponsive. You are trying to feel small amounts of ankle pressure through them, and a shoe sole is exactly the wrong thing in the way.
One candidate felt the sensitivity shifting mid-exercise and couldn’t tell whether it was a deliberate mechanic or a fault.
Never resolved. If it happens to you, keep flying it. Don’t try to diagnose it mid-run.
There is a practice version on this site. It runs on a joystick, a mouse, the keyboard or a touch pad, and rudder pedals can take the sideways axis like the real thing.
It is my own version rather than the real subtest. What it trains is the thing the test is actually about: your input sets how fast the dot moves rather than where it goes, so you are always holding a correction rather than dragging the dot back and letting go.
What the test isA grid of tiles, each holding either a red letter or a unique symbol, with a small black number in the bottom corner. A target tile is shown at the bottom of the screen with its number replaced by question marks; you find the matching tile in the grid and give the number that belongs to it. It sits with Target Recognition and Table Reading as the rapid visual-scanning family, so practice here pays off three times.
Format description from the published CBAT TMI test guides
(rafcbat.wordpress.com) and the Air Defence Academy test guides
(airdefenceacademy.com/testguides).
Straightforward enough in principle, with one quirk in some practice apps that’s worth knowing about.
In the real test the symbol to number mapping mostly holds still, but every ten questions or so it adds new symbols and shuffles the numbers attached to them.
Accounts differ, and some people think it changes every time. Safest approach is to use the mapping while it holds, then re-check it whenever new symbols appear, instead of trusting your memory across the whole test.
In some practice apps the mapping never changes within a test, so you can memorise it and skip the actual skill.
If your tool does that, avoid the shortcut on purpose. Scoring well by learning a fixed key is training you for a test that doesn’t exist.
Both common versions of visual search show up in the real test, and the number of items you’re scanning keeps growing.
The skill carries into the table reading and vigilance tests as well, so time here isn’t narrowly spent.
What the test isTwo tests. The first shows 3D shapes marked with black dots as reference points, to be mentally rotated and matched against the options underneath, and only one is correct. The second shows several flat shapes whose edges are labelled with letters marking where they join (the edge labelled b meets the other edge labelled b), and asks which of the assembled figures at the bottom is right once everything is fitted together.
Format description from the published CBAT TMI test guides
(rafcbat.wordpress.com) and the Air Defence Academy test guides
(airdefenceacademy.com/testguides).
Two spatial tests: rotate the solid, or fit the labelled pieces together. Drill the skill and the exact format stops mattering.
These two didn’t appear in the official 2018 CBAT information, so the description above comes from the published guides rather than from a candidate write-up.
Worth knowing if you’re comparing notes with someone who sat an older battery and remembers it differently. The underlying skill is the same either way, holding a shape in your head and turning it, so practise that rather than memorising a layout.
The only usable signal is that heavy flight sim experience is said to help with 3D visualisation work.
Thin, but it’s the only thing here I’d stand behind.
What the test isThree components running at once. Colours: red, yellow and green dots travel horizontally toward colour-coded zones and you press the matching key at the moment of contact. Letters: a 4–6 letter sequence shows for about five seconds, then vanishes, and you pick it out of four scrambled options in the screen corners. Numbers: rapid arithmetic typed in before the timer moves on, with no review state and no way back to a previous answer.
Format description from the published CBAT TMI test guides
(rafcbat.wordpress.com) and the Air Defence Academy test guides
(airdefenceacademy.com/testguides).
Published as replaced by FLAG in the RAF battery back in 2021, but still live elsewhere. Check which one you’re sitting before you spend a week on it.
If you’re sitting the current RAF CBAT, you almost certainly get FLAG instead of CLAN. Nobody sitting the RAF version reports getting CLAN.
Graded red because practising CLAN for an RAF CBAT looks like wasted effort. It also explains a common source of confusion: some practice apps carry CLAN and others FLAG, and people compare scores across them without realising they’re preparing for different tests. SkyWatch has both, so pick the one your battery uses.
At least one overseas candidate sat CLAN and got no FLAG at all, and a New Zealand candidate reported CLAN still being in their battery.
An Australian candidate was told by the psychologist running their testing that the same software is used by the UK, Canada, New Zealand and Australia, and that each military picks its own tests from it, so CLAN versus FLAG is a choice each service makes. So if you’re applying outside the RAF, don’t assume the FLAG advice higher up this page applies to you. Ask what your battery contains before you plan around it. The Royal Navy has gone the other way and moved to FLAG, so a Navy guidance document still showing CLAN is worth checking rather than trusting.
It’s coloured diamonds, maths and remembering codes, structurally very close to FLAG.
Which is good news if you’ve been drilling FLAG and then find out you’re sitting this. The three concurrent tasks map across almost one for one, and the same advice applies: bank the maths, don’t defend the memory task to the death.
What the test isTwo separate visual-estimation sub-tests, both multiple choice. Angles: two lines form an angle on screen and you identify its value from the options; the angle rotates and the options narrow as it goes, so there’s nothing to anchor to. Bearings: a 360° compass circle carries lettered points (A, B, C…) plotted across it, and you estimate a named point’s bearing clockwise from north, again picking the closest option as they converge.
Format description from the published CBAT TMI test guides
(rafcbat.wordpress.com) and the Air Defence Academy test guides
(airdefenceacademy.com/testguides).
One of the few tests where raw repetition is genuinely the whole answer, and one of the easiest to practise for free.
On bearings, anchor to the four cardinal points first and count from the nearest one to the target rather than estimating from north every time.
This is from the published guides rather than from anyone who sat it, but it’s sound and it costs nothing to adopt. It cuts a good chunk off your estimation time, which matters because the multiple-choice options get closer together as you go.
Angles is pure speed and accuracy built through repetition. There’s no clever method to find.
Free angle-estimation and protractor tools are all over the web and they’re a decent use of a spare twenty minutes a day. This is the least mysterious test on the battery. You just have to put the reps in.
Almost no first-hand information exists on this test anywhere.
Nobody I found had described sitting it. That’s not a reason to worry, given how simple the format is, but treat any confident account of the exact question count or timing with suspicion.
Angles is well worth practising, and one candidate noted the real screen is laid out its own way.
So train the skill, estimating an angle quickly, rather than getting used to one particular layout.
What the test isA single horizontal string of digits appears for exactly five seconds, then disappears and is replaced by one multiple-choice question: how many times did a given digit appear. The format never varies: always the frequency of one digit, always five seconds, always multiple choice. Only the string length grows, from around five digits to around fifteen.
Format description from the published CBAT TMI test guides
(rafcbat.wordpress.com) and the Air Defence Academy test guides
(airdefenceacademy.com/testguides).
Mechanically the simplest thing on the battery, and the one where knowing the format in advance genuinely changes how you play it.
Difficulty comes from string length alone. Nothing else changes between the first question and the last.
That means the right strategy changes partway through. Short strings you can memorise whole and count afterwards. Once you’re past ten digits or so you have to track frequencies live while the string is still on screen, because there won’t be anything to recall.
Track the high-frequency digits in the closing seconds, not the rare ones.
A digit that appeared once or twice you can often recover by elimination. One that appeared four or five times you can’t. Spend the last second of the display window on whatever’s repeating.
Nothing is scored on how fast you answer afterwards. The whole test lives inside the five-second display window.
So don’t rush the answer. Once the string is gone, take the time to reconstruct properly. There’s no clock pressure left to react to.
Listed as not included in the Royal Navy FAT version.
One of three tests dropped from the Navy version, along with Dynamic Projection and Numerical Operations. Worth checking against your own service before you build a schedule around it.
What the test isA physical double-sided reference sheet worked against on-screen prompts, in two parts. Part one gives two values and a coordinate grid running −17 to +17 on both axes; you bring one value across and one down and read the intersection, which is the same whichever axis you assign each number to. Part two puts Air Speed, Wind Velocity and Wind Angle on screen and needs a three-step lookup: table by air speed, row by wind velocity, then column by wind angle under either Drift Correction or Ground Speed. Five numbered options each time.
Format description from the published CBAT TMI test guides
(rafcbat.wordpress.com) and the Air Defence Academy test guides
(airdefenceacademy.com/testguides).
The only test worked against a paper sheet rather than entirely on screen, which is a bigger adjustment than it sounds on a day that otherwise bans paper.
Visual search practice is said to carry directly into this test.
From someone who passed, and it’s a good reason to treat visual search practice as three tests’ worth of value rather than one.
The reference sheet is pre-printed and laminated, and it is the only physical apparatus you get all day.
Confirmed by a candidate who noted it as the single exception to the no-paper rule. Worth knowing so you aren’t thrown when something lands on the desk after a morning of being told you can’t have anything on it. It also means the sheet is a fixed, wipe-clean object you slide about rather than something you can mark up.
You can build your own practice for this one in a couple of minutes. Draw out a grid running to 17 in each direction, write yourself some coordinate pairs, and drill the lookups.
Recommended by someone who had just sat it, and it’s the only concrete prep anyone has offered for MATF. Do it on a printed sheet next to your screen rather than all on one display, because moving between two surfaces is the part that actually costs time.
On the coordinate grid, the intersection value is the same whichever way round you assign the two numbers to the axes.
So don’t waste any time working out which value belongs on which axis. Bring one across, one down, read where they meet. That decision is a free half-second on every question in part one.
Reading how a grid lookup works is easy. Running it fast and accurately while managing a physical sheet and a screen at the same time is the actual test.
If you practise this at all, practise it with a printed sheet beside you rather than everything on one display. The hand movement and the eye movement between two surfaces is the bit that costs you.
What the test isA joystick, a first-person aircraft camera feed and a crosshair fixed at the centre of the display. Targets appear at varying distances and speeds, and each has to be photographed three times with the trigger before it disappears or is replaced, and a replacement resets the capture count to zero. Targets pass behind obstructions, so reacquiring them means keeping the camera moving in the direction and at roughly the speed the target was travelling while it’s out of sight.
Format description from the published CBAT TMI test guides
(rafcbat.wordpress.com) and the Air Defence Academy test guides
(airdefenceacademy.com/testguides).
Built to overload you. What’s measured is where your tracking degrades, not whether it ever does, so don’t read the last thirty seconds as a verdict on how you did.
This test on its own can cost you the role. One candidate finished eight points under the pilot mark and named this as the test he failed.
Worth knowing because it’s easy to write this one off as unpreparable and therefore not worth thinking about. You can’t train the hardware, but you can arrive knowing to calibrate on the opening targets rather than lunging at the first one, and that alone is worth marks.
Don’t try to snap the crosshair onto a target in one big joystick movement. Close the distance with a coarse input, then fine-adjust before you trigger.
The single most useful thing in the published guidance on this test. Big corrections overshoot, and every overshoot costs you more than the slower approach would have.
Joystick sensitivity differs from station to station, so the opening moments are effectively yours to calibrate with.
Make small deliberate inputs at the start and watch how far the camera actually moves before you commit to chasing anything. Going in assuming your practice sensitivity carries over is how the first target gets away from you.
When a target goes behind something, keep the camera moving in the direction and at the speed it was travelling rather than stopping to wait.
Stopping means reacquiring from scratch when it reappears. Tracking the gap keeps you roughly on it.
This runs on the same joystick hardware as the motor test, where the stations are a documented variable: different dead zones, sensitivities and wear.
Which means the same rule applies: if your station feels genuinely unusable, say so at the start rather than fighting it. Staff keep spares. That’s reported from the motor test rather than this one, but it’s the same room and the same kit.
The practice run ends on its own and the scored test starts straight away, with no break in between.
One candidate missed a role because of this and said several others in the room hadn’t noticed the switch either. Fly the practice as if it already counts.
The camera is mounted on an aircraft moving through a large scene rather than sitting in one place, and there are a lot of targets to photograph.
Several recent candidates described it this way. Expect the view to move on its own while you aim.
You have to keep each target inside the box the whole time, not just line up and wait for it to come out from cover.
From a candidate who scored very well. Targets sometimes turn too tightly to hold perfectly, and that is fine; pick them back up and carry on.
It is long and constant, and candidates reported arm fatigue and hand cramp towards the end.
Hold the stick loosely and rest your forearm where you can. If you practise, do full-length runs rather than short bursts.
The stick works like an aircraft’s: push forward to move the view down, pull back to move it up.
From one detailed account. If your practice is set up the other way round, switch it before the day.
What the test isUp to about thirteen pages of information on one broad subject (past topics include rocks, insects and plants, and renewable energy) with only two pages readable at a time. The questions are multiple choice and are deliberately not answered plainly anywhere in the text: you locate the relevant pieces across separate tabs and infer from them. Usually three parts of increasing size, roughly 15–20, 20–25 and 25–30 minutes, with around three minutes of familiarisation up front.
Format description from the published CBAT TMI test guides
(rafcbat.wordpress.com) and the Air Defence Academy test guides
(airdefenceacademy.com/testguides).
Structurally this is the System Logic Test with prose instead of figures, so prepare for them together.
No subject knowledge is required. Whatever the topic turns out to be, everything you need is in the tabs.
Worth internalising, because the topic can look intimidating. You’re not being tested on rocks or renewable energy. You’re being tested on whether you can join two pieces of information from different pages.
The answers aren’t written plainly in the text. You have to join the dots between separate sections.
So don’t keyword-hunt for a phrase from the question. If you find your answer sitting there in one sentence, check it, because that’s usually the distractor rather than the answer.
One candidate who sat it describes it plainly: a table of eight tabs of information, three minutes to read them, then the questions start.
The first usable first-hand account of this test I’ve found. Eight tabs is fewer than the published guides imply, so don’t brace for a textbook. Note also that he treated it as a scanning job, reading the question and then going to find the pieces, rather than trying to hold the content in his head.
Roughly three minutes per question, and it moves on by itself when the time is up.
From the same account. The auto-advance is the part worth planning for: on a question you can’t place, you are not choosing between answering and moving on. It will move on regardless, so put something down early and refine it rather than running the clock out with nothing selected.
The published guides say you can skip questions and come back, and you get about three minutes up front to familiarise yourself.
The three minutes up front matches the first-hand account. Skipping and returning doesn’t sit easily with a question timer that advances on its own, so check how the two work together during your reading time. Either way, use the opening three minutes to learn where things are rather than to read for content, the same approach as the System Logic Test.
What the test isA 9×9 grid filling with stars, cleared by typing coordinate pairs: row number first, then column, so a star at 2,7 is entered as those two digits. Two task types run: yellow priority tasks that appear during the test and carry bonus points if cleared quickly, and the routine task of cancelling every remaining star. The screen gets progressively more cluttered as it goes.
Format description from the published CBAT TMI test guides
(rafcbat.wordpress.com) and the Air Defence Academy test guides
(airdefenceacademy.com/testguides).
Mechanically the simplest test on the battery, which is exactly the point. It measures whether you can hold attention on something dull, not whether you can do it.
Visual search practice is said to help here as well as on table reading.
The only thing anyone with experience has said about this test. Another reason visual search is the best-value thing to practise on the whole battery.
Work along a row or column in sequence (2,1 then 2,2 then 2,3) rather than jumping around the grid.
You stop re-reading the labels for every entry, which is where the time actually goes.
Edge squares can be entered without checking the grid labels at all, so they’re quicker than anything in the middle.
Clear the edges when you’re under pressure and leave the interior for when you’ve got breathing room.
Yellow priority tasks appear mid-test and carry bonus points if you deal with them quickly.
So the routine star-clearing is the background job, not the main one. Break off for a priority task the moment one appears.
Use the practice session to drill the coordinate entry itself.
The bottleneck is the keying, not the finding. If typing a pair of digits is automatic before the test starts, you’ve already got most of what this test is measuring.
Expect the grid to fill with far more stars than you can clear. Several candidates described it as close to impossible to empty.
One candidate never got near emptying the board and still scored full marks. Clearing the grid is not the goal. Work steadily and catch the priority stars.
Priority stars need an extra key press before you type the coordinate.
Two candidates described this. Practise the two-step entry until it is automatic, because you will be doing it when the grid is at its busiest.
Test day
What the day is actually like
The tests are only half of it. Here’s how the day itself runs, from the kit and the rules to the bits of admin nobody warns you about.
What the day actually feels like
It’s far more relaxed than almost anyone expects. Several people barely saw an invigilator all day.
One candidate went looking for the invigilator mid-test, couldn’t find him in the hall or the break room, went back to his seat and carried on. Coats and phones usually go on a desk in the middle of the room, your bag having been left in the cloakroom. Don’t expect an exam hall.
No pen, no paper, no ruler, at any point, for any test. It’s all on screen and in your head.
This is the one thing staff actually police, and it’s specifically about nothing on the desk you could work something out with. It’s also why the estimation and mental arithmetic techniques in this guide matter as much as they do.
Three pieces of kit at every booth: an Elgato Stream Deck XL in place of a keyboard, an Airbus-branded joystick, and rudder pedals.
The Stream Deck is roughly half the length of a normal keyboard with a 3×3 number pad, and it has been named by candidates specifically enough that you can go and look at a picture of one. You’ll fumble the first couple of presses and then get used to it. People rate it well above the older kit.
You’re allowed to move the kit around on the desk. On a joystick-only test, push the Stream Deck out of the way and fly two-handed.
From someone who wished he’d worked it out sooner. Nothing is bolted down and nobody minds. It matters most on the tracking test, where people consistently report their arm aching by the end, and two hands on the stick is a great deal steadier than one.
You can’t read anything out loud, but mouthing the words silently works fine and does help.
If talking yourself through a problem is part of how you think, that’s the workaround people use.
Most of the day is one-handed, because your other hand is usually busy. Both hands only really come in for the quick arithmetic test.
Worth practising one-handed if you’ve been using two.
It’s eight to ten hours in front of a screen. The eye strain is real and it arrives long before the day ends.
Two separate accounts describe not being able to focus properly afterwards, one sitting in the car park for ten minutes before driving. On your breaks, look out of a window at something distant and stay off your phone. A screen break that is still a screen isn’t a break. Plan your journey home around being genuinely tired.
A short break comes round every thirty to sixty minutes (roughly every one to three tasks) plus about thirty minutes for lunch. Five minutes is the recommendation, not a rule; you can take longer and the only cost is finishing later.
Accounts vary on the exact interval, which suggests it follows the tasks rather than the clock. The break room has a screen showing your predicted finish time and how much of your break is left, and the prediction gets more accurate as the day goes on. One candidate had the latest finish in the room by about an hour, took every break anyway, and passed comfortably. Nobody is scoring you on finishing early.
You sit in an individual booth, much like a driving theory test. The room holds around 45 but is often nowhere near full.
One sitting had 33 candidates. Everyone works through the tests in the same order, which is why your break-room conversations are with people who have just done exactly what you have just done.
You can’t leave the test room except at a break.
Which is really a hydration point. Drink at every break rather than steadily through a task, or you’ll spend an exercise thinking about something other than the exercise.
You can read the instructions for each test twice, and most tests give you a practice run you can take twice as well. The practice uses the same questions both times.
This is free and a lot of people leave it on the table. Because the questions repeat, the second go isn’t about getting them right. It’s about getting the mechanics automatic. It pays off most on the joystick and rudder tasks, where the whole point is to find out how sensitive your particular station is before anything is being scored.
Your recruiter sends you a pack setting out every test you’ll sit. Read it properly.
It comes up again and again from people who passed, usually in the form of a regret. One described the pack as a godsend and then admitted he hadn’t worked out how a test behaved until it was over. It tells you which tests apply to your role, which is the single most useful thing to know when you’re planning what to practise, and it costs nothing because they’ve already sent it to you.
What everyone who passed tells you to do
Read the instructions fully, twice, and use every second of the practice run. Take all the time they give you.
The most repeated advice from people who passed, and it’s free marks. Nobody scores you for finishing early, and people finishing ahead of you means nothing at all. On DPT especially, go into the practice attempt with a specific question you want answered.
Use every break. They come round every one to three tasks. Drink water, eat something, and don’t replay the test you’ve just done.
Chewing over a task you’ve already submitted costs you the next one and gains you nothing. It’s the most common way one bad exercise turns into a bad day.
Protect your sleep. The rooms are decent but they run hot, mattresses can be springy, and it’s nerves that actually keep people awake.
Open the windows before you head to dinner. A sleeping bag comes up again and again. Some people bring melatonin, at least one brings a plug-in fan. The candidate who lost a night to nerves blames it directly for his failure, so this is a real performance issue and not fussiness.
Bring your own pillow. The ones provided come up as the single most complained-about thing in the block.
Independently recommended by several people who passed. Two other cheap additions from the same accounts: an eye mask, because you may want to be asleep at half eight while it’s still light, and earplugs, because the doors in the accommodation slam. You are trying to buy one decent night before a ten-hour day, and this is the cheapest way to do it.
Your bag stays named and in the cloakroom in the candidates’ mess. It does not come into the test room.
Labels are at the candidates’ mess entrance and staff will point you at them, so label the bag and leave it there. Older accounts of people carrying a bag into the testing room are out of date, so plan on being separated from it for the day and take what you actually need in your pockets.
Drinks go on the floor, and you generally keep your phone on you.
So a bottle and a snack are worth having on you. Nothing else needs to come in with you.
You report to the main guard room early, around 06:45, to pick up your site pass. Most people finish somewhere between 2pm and 4pm.
It’s a long day. Plan your travel and food around a mid-afternoon finish, not an early one.
On the train, Grantham is the station people travel via.
Mentioned in passing by a candidate who also warned against buying an energy drink there to get himself going, because he crashed later in the day. Worth planning the last leg from the station in advance, given you need to be at the guard room before seven the following morning.
Turn up smart. A collared shirt and trousers is the normal standard, and the pack you’re sent sets it out. Several accounts describe wearing shirt and tie for the whole time they were at Cranwell, dinner the night before included.
There is a persistent line that CBAT is only an aptitude test so nobody cares what you wear. Don’t rely on it: candidates who turned up in a tracksuit got noticed for it, and one of the more detailed write-ups makes the point that you are on an RAF station being looked at by RAF staff from the moment you arrive. Within smart, still optimise for sitting in a chair for ten hours: you can loosen the tie and take the jacket off once you’re in the booth, and you should wear nothing that gets in the way of working the controls.
Take your shoes off for the rudder pedals.
Reported by someone who passed, and it costs you nothing. The pedals are worked with ankle pressure and the accounts consistently describe them as stiff and low on sensitivity, so anything between your foot and the pedal is working against you.
If you drive, you can leave things in the car rather than carrying them all day.
Several people recommend it. Anything you do bring goes in the cloakroom under your name, so the car saves you a trip back to the mess. Take water and snacks in with you.
Candidate accommodation has moved out of the old mess into a block, and the rooms are not en-suite.
Confirmed separately by more than one person who has stayed in the block, and worth flagging because the published guidance hasn’t been updated and still describes the old arrangement. Showers and toilets are shared down the corridor, so bring something to walk down there in. Rooms run hot, which everyone mentions.
Scoring, scheduling and how long a pass lasts
The pass mark for pilot is 112, out of a maximum of 180, and it is the figure candidates quote off their own results sheets more consistently than any other.
This is the single most repeated correction I found. A serving RAF pilot posted specifically to clear it up, and candidate after candidate quotes 112 against their own result: 95, 104 and 105 as failures, 116, 117, 120, 125, 134 and 161 as passes. Treat it as the number in use rather than a permanent one; see the history below.
Other roles sit lower, and the figure is printed on your own results sheet against each one. Reported marks: weapons systems officer 100, weapons systems operator around 90 to 108 depending on the stream, air operations (systems) officer 90, intelligence officer around 95.
All quoted by candidates reading their own sheets, including one who got 111 for a WSO cutoff of 100 and another who missed WSOp ISR on 80 against 90 while clearing operations officer on the same day’s work. The spread is the point: a score that fails you for one role passes you comfortably for another, which is why so many people walk out qualified for something they hadn’t considered. Don’t assume the pilot figure applies to you.
The pilot cutoff has moved before and can move again. The published guides carried 128 for years; recent results sheets show 112; one candidate who sat it more than once recalls it going 112, then 120, then back to 112.
Worth setting out plainly because the stale figures are still circulating and people talk themselves out of applying over them. A candidate who passed compared his own results sheet against the published 128 and found it simply out of date. The honest position is that the RAF sets this and can change it whenever it likes, so treat any number you read anywhere, including here, as the last one somebody reported rather than a guarantee, and read your own sheet on the day.
A pass is a pass, and the pass mark is the only threshold you have to clear. Separately, the average scores of pilot applicants have been published, and they sit well above it.
A Ministry of Defence freedom of information release put the average CBAT score for pilot applicants around 136, with those actually allocated a place a little higher; other figures quoted from the same family of releases run to roughly 145, and around 167 some years earlier. One serving source describes 130 to 140 as typical for successful intake. Set against that, a serving pilot was blunt that "competitive" is a legacy term from when there was a large sift, people are on record passing on 112 and 114 and going into fast jet training, and the figures above were gathered while the system was unwinding from a long period with pilot applications closed, which inflates them. Both things are in the record, so both are here. What nobody disputes is that clearing the mark is what puts you through, and the selection board carries more weight than the number.
Each test is marked 1 to 9, a stanine. Those feed weighted domains, and the domain totals produce the overall score for each role.
Candidates use the word stanine constantly once they’ve had their results, and describe missing out "by one stanine" on a single domain. This is why one weak test can matter out of proportion to its share of the day.
Clearing the overall score is not enough on its own. Each domain carries its own minimum, and falling one stanine short of any of them fails you for that role even with a comfortable total.
This is the most consequential mechanic in the whole assessment and it catches people constantly. One candidate cleared the overall mark at 101 against 90 and still lost the role on spatial reasoning. Another scored 117 against a 90 requirement and still saw red against the role. A third passed everything except intelligence, missed by a single point on FLAG, and lost the three aircrew roles he actually wanted. Do not write off any single test on the basis that your total will carry it.
The results page lists your domains with a weighting against each. One candidate’s, for intelligence officer, read: Perception ×10 (table reading, target recognition), short-term memory ×13 (situational awareness), spatial reasoning ×25 (directions and distances, spatial integration), symbolic reasoning ×41 (system logic, verbal logic) and central information processing ×11 (FLAG).
From two candidates quoting their own results pages, and the domains they list differ. The other, for a different role, had strategic task management covering the multitasking and auditory tests, perception covering target recognition and vigilance, and symbolic reasoning covering situational awareness and the numerical test. So both which domains apply and which tests feed them depend on the role you’re sitting for. Useful for working out where your effort pays, but don’t treat one candidate’s table as the universal scheme.
Some tests are sat more than once. The published pilot rubric scores the multitasking test three times, the numerical test three times and the auditory test twice.
Thirteen scored tests, but more than thirteen sittings. It’s another reason the tests people rate as most tiring are the ones that come round again.
Pilot and weapons systems operator candidates sit the entire battery. Other roles sit only the tests their role needs, plus any test shared with a role they’re also being scored for.
Which means going in for pilot leaves every other aircrew role open to you on the same day’s results, and it’s why so many people who fail pilot walk out qualified for something else they hadn’t considered.
You get your results the same day. A boarding officer takes you through a booklet role by role, colour coded, usually within minutes of you finishing.
One candidate’s officer hadn’t even read it himself yet and opened with "lots of green, well done". Green is a clear pass, red is a fail, and yellow appears to be a pass you scraped. It isn’t especially private, so expect to take the news standing at a desk while other people finish.
CBAT isn’t limited to Tuesdays and Thursdays. It runs on other days too, and it overlaps with selection board dates.
Some recruiters still quote the Tuesday and Thursday rule. One candidate sat his on a Wednesday alongside about thirty others, and a call to the recruitment office confirmed other days do run occasionally. Worth checking rather than assuming.
A CBAT pass is valid for three years, and it does expire. One candidate’s ran out almost exactly three years after he passed, and his appeal for an extension was turned down.
The three-year figure is stated flatly and repeatedly, including by people advising others to get the aptitude test done early precisely because the result banks for that long. It cuts both ways: it means you can sit CBAT while still at university and turn up to the rest of the process with a score already on the door, and it means a long wait at any later stage can quietly run your result out. If there’s any chance of a gap between passing CBAT and getting to the selection board, find out your expiry date early.
The wait for a retake is around a year, and staff will warn you that most people don’t improve much. The accounts bear that out more often than not.
One candidate sat it twice and got the exact same scores both times, and says plainly he regrets the year he spent on it. Against that, one who ignored the same warning, went to university and flew with a University Air Squadron in between came back about 14 points up, then another 13 on a third attempt. The pattern seems to be that a retake on its own changes little, but a retake after a couple of years of genuinely relevant flying and study can. Aptitude is also reckoned to rise into your early twenties and tail off after your mid-twenties.
Your score isn’t used to stream you once you’re in. Passing is what counts.
From the same serving pilot, stated flatly: you will never be held against that score after recruitment, and it has no influence on streaming. That contradicts a persistent belief that a higher score buys you fast jets. One well-placed source rather than several, but he’s the best-placed source I found on it.
The psychology
How people think they did versus how they actually did
This is the most consistent pattern I came across anywhere, and the one I’d most want you to take in before test day. Of the candidates who described both how a test felt at the time and what they actually scored, 5 out of 7 did a lot better than they thought. The only two whose pessimism turned out to be right were a candidate who’d lost a night’s sleep, and one who dropped a single mark to a misread word.
5/7
did better than
they thought they had
they thought they had
3
passers who nearly wrote
themselves off mid-test
themselves off mid-test
1
passed everything after
two previous failures
two previous failures
0
people whose panic on one
test predicted the outcome
test predicted the outcome
| Candidate | How it felt at the time | What actually happened | Gap |
|---|---|---|---|
| A candidate going for air traffic control | Reckoned he’d "bottled it" on the bearings test at the halfway point, and seriously wondered whether it was worth staying to finish the day. | Passed for air traffic control and weapons controller, which were the roles he wanted. | ✓ Did better than they felt |
| A candidate going for weapons control | Felt awful about the bearings test. Got reset to the gate several times and missed an intercept completely. | Scored well on that same test, passed, and went on to pass the selection board. | ✓ Did better than they felt |
| A third-time candidate | Had failed everything twice already and expected the same again. | Passed everything, and by his own account not narrowly. | ✓ Did better than they felt |
| A candidate who applied as WSO | Applied for weapons systems officer because he didn’t think he was good enough for pilot. | Scored higher for pilot than he did for WSO. | ✓ Did better than they felt |
| A high-scoring pilot candidate | Expected the real tests to be harder than practice. | Passed with one of the highest scores I’ve seen reported, and found the real thing technically easier in a lot of ways. | ✓ Did better than they felt |
| A candidate on an overseas two-day version | Knew he was struggling after a sleepless night before day two. | Didn’t pass for pilot. The one case where the bad feeling turned out to be right. | ! Feeling was accurate |
| A candidate drilling instruments | Thought the elimination method had made instruments watertight. | 14 out of 15, losing one to reading "descending" as "ascending". | ! Feeling was accurate |
So what do you do with that? How you feel about a test you’ve just
finished tells you nothing useful, so don’t let it bleed into the next one. The bloke who seriously
thought about leaving at the halfway mark passed for both roles he wanted. Finish the day, whatever
happens in the middle of it.
One caveat on all this optimism. People who pass tend to write up their
experience. People who don’t usually go quiet. So the accounts that exist, including the ones behind this
guide, lean heavily towards success. The "it felt worse than it scored" pattern is real and worth
trusting. Any sense you get of an easy pass rate is not.
Preparation
What actually helps
Ranked by how often people who passed give each one credit. The bar shows weight of recommendation, not measured effectiveness, because nobody has run a proper comparison.
A dedicated CBAT practice tool, like SkyWatch
Credited far more often than anything else. The reason people give is usually interface familiarity and stamina rather than the content itself. Knowing where to look, and being able to hold concentration for the length of a task. Nearly everyone who passed and talked about their prep had used one heavily.
Mental arithmetic drilled separately
Times tables, division, and speed distance time worked out by hand, away from any app. This is the one thing people say the apps don’t replace, and the candidate who turned two failures into a clean pass was blunt that an app alone hadn’t been enough.
The official test guides
Best for working out what each test actually is before you try to get good at it. Some of the practice tools linked from them are a lot harder than the real test, which is fine for overtraining but misleading if you’re trying to judge whether you’re ready.
Candidate write-ups and discussion
This is where the corrections live. Nearly every difference between the practice tools and the real test in this guide came from someone who’d just sat it, not from anything published. If you read nothing else, read recent first-hand accounts.
Flight simulators
Said to help with 3D visualisation and reading instruments. Nobody reports any benefit to the memory, arithmetic or multitasking tests. Enjoyable, but not an efficient route unless you already play.
A joystick, genuinely disputed
Worth setting out both sides, because people spend real money on this one. Against: a candidate who scored 158 for pilot was emphatic that buying a stick for CBAT is a waste of time and money, on the grounds that the joystick tasks are simple, the station sensitivity is nothing like yours, and you get practice runs on the day anyway. He rated ordinary gaming just as useful for the hand-eye side. For: several people who passed credit sim and stick time for the tracking and trace tests specifically, and one described the joystick exercises as the part he enjoyed because they felt like a game he already knew. The reasonable middle is that a stick is not something to buy for the test, but if you already own one it is worth using. Of the tests on this site, the auditory capacity test and the rapid tracking test are the two that take stick input.
General practice websites
Speed distance time drills, instrument comprehension sets, angle estimation tools. Usually harder than the real test, so use them as conditioning and not as a benchmark.
What a preparation that works looks like. SkyWatch or a similar tool for
interface familiarity and stamina. Mental arithmetic drilled separately, by hand. The official guides so you know
what each test is. And recent first-hand accounts for all the corrections that none of the published
material carries. The candidate who failed twice and then passed everything did all four, and said
plainly that leaning on an app alone had been his mistake.
Other services
The Navy, Australian and Canadian equivalents
If you’re sitting another service’s aptitude test, most of this guide still applies. Here’s what’s known to be different.
Royal Navy, Fleet Air Arm
Your CBAT result carries across to the Royal Navy. One candidate sat it for the RAF and had the same result count for a Navy pilot application.
That makes a single sitting a genuine second route rather than something you have to repeat, and it means applying to both services at once is a sensible hedge rather than a duplication of effort. A shared aptitude result wouldn’t automatically extend to medicals though, which are a separate question entirely.
The Navy pass mark is reportedly higher than the RAF’s, and there are far fewer places.
One candidate was told by his recruiter that the Navy mark sits above the RAF’s, and reports the Navy taking on three pilots in a calendar year with the lowest successful CBAT score around 165. So "it carries across" is true but it doesn’t make the Navy the easier option. On these numbers it’s considerably harder.
The Navy pilot age limit runs higher than the RAF one, reportedly up to 26.
Relevant if you’re near the RAF age ceiling and worried about getting through the remaining stages in time.
The Navy’s Observer role gets described as their version of a weapons systems officer, with navigation on top.
It covers weapons and sonar, mission command such as directing ships and aircraft, and navigation. The Observer is often the most senior aircrew on board.
Australia, ADF
The Australian test, MACTS, reportedly runs on the same software as the RAF CBAT. Each country’s military picks which tests from it to use.
A 2026 Australian candidate was told this by the psychologist running the testing at their Aviation Screening Program, and named the UK, Canada, New Zealand and Australia as the countries using it. It explains why candidates have long described the two as near-identical, and why the one clear difference is a swapped test rather than a different battery. One account, passed on second-hand, so check it if it matters to you.
The Australian test uses CLAN instead of FLAG.
Several Australian candidates name CLAN, and none describes sitting FLAG. CLAN is the older sibling of FLAG covered further up this page. If you’re preparing for the Australian version, that’s the big structural difference to plan around. There is a separate SkyWatch guide to MACTS for Australian applicants.
Everything else lines up closely. Instruments, visual search, situational awareness, directions and distances and the trace tests all turn up in both.
So most UK-focused prep carries over directly. Don’t throw this guide out over one test.
Australia reportedly allows only two attempts, on the reasoning that people don’t improve enough after two.
Striking, given a UK candidate in this guide passed everything on his third go after two comprehensive failures. Where you get more than one attempt, take it.
Canada, RCAF
Canada runs it over two days, with the joystick and pedal work on day two.
That makes the first night’s sleep matter in a way the UK single day doesn’t. The one detailed account I have is of exactly that going wrong.
Both trace tests and a sensory motor test appear in the Canadian version too.
So the trace and motor advice in this guide applies.
The short version. The Royal Navy is said to run the same test under a
different name, with the result recognised by the RAF. That could be a useful second route, but it rests
on one account, so check it with a recruiter. The Australian version reportedly runs on the same software,
with each country picking its own tests, which is why it has CLAN where the RAF has FLAG. Canada splits it over two days, which puts a lot of weight on the first night’s sleep.