How to prepare for the TARA

There is no syllabus to revise, so TARA preparation is not revision in the usual sense.

What you are training is how fast you recognise a question type and how cleanly you reason under a clock. Both of those move with practice.

Neither moves with cramming.

The plan, in one paragraph, before the detail:

Start about twelve weeks out. Learn the question types untimed, drill them against the clock, sit full papers under real conditions, and only at the very end sit the two official TARA tests as your two dress rehearsals.

Do as much of that as you can on questions written to the TARA format, whether that is the free TSA and BMAT archive or a dedicated bank. The closer your practice is to the real thing, the more your practice scores mean.

Ours is one option and we are obviously not neutral about it, so try the free tier and judge the questions yourself.

The sequence matters more than the hours, and it matters for a reason specific to this test.

TARA is one year old, and there are two official tests in existence. That is the entire supply.

So the central problem of preparing for it is not finding material. It is not wasting the only material that is genuinely TARA.

That last number is the one to hold on to, and it is covered in full further down this page.

If you have not worked out yet whether you even need to sit TARA, start with the complete TARA guide instead.

Four facts about the format change how you should prepare. Not one of them is about content.

Here's how that breaks down:

The modules run in a fixed order and you cannot go back. Finish Critical Thinking badly and you carry it into Problem Solving with no way to return.

So practicing a module in isolation is fine early on, and useless as a final rehearsal. The last few sittings need to run 120 minutes end to end, in order.

The two scores are reported separately and there is no total. A strong Critical Thinking score cannot rescue a weak Problem Solving one, because no university ever sees them averaged.

Which means preparing only for the module you enjoy is the single most expensive choice available to you. Split your time towards the weaker one, not the stronger.

There is no negative marking. A wrong answer and a blank are worth exactly the same, so an unanswered question is a wasted lottery ticket rather than a safe choice.

You sit it once. One sitting per admissions cycle, no resit, and Oxford applicants must use the October date.

There is no second attempt to average against.

The Writing Task is unmarked, which is why almost everyone ignores it. It is also sent as text to every university you applied to, which is why almost everyone is wrong.

This is where most people get stuck, and the confusion comes from an official line that is entirely true and widely misread.

UCL's own Computer Science admissions page states that "the TARA exam does not require any subject knowledge so applicants do not have to conduct additional revision for the exam".

UAT-UK say something similar:

"all necessary preparatory materials are available, free of charge, to all candidates".

Both are accurate. Neither says preparation does not work.

What they are telling you is that there is nothing to learn. No formulae, no set texts, no topic list to tick off.

You cannot revise for TARA the way you revise for a Chemistry paper, because there is no body of knowledge being examined.

What is being examined is a set of habits, and habits are trainable.

Recognition speed. Every Critical Thinking question is one of seven types, and every Problem Solving question is one of three skills.

Knowing which one you are looking at within a few seconds is worth more than any single technique, because it tells you what to do next.

Resistance to the wrong options. On the real papers every wrong option is engineered to catch a specific, predictable misreading.

None of them are filler. Learning the common traps is learning most of the test.

Pace under pressure. This is the one the data says is binding, and it gets its own section below.

There is a limit worth stating in the same breath, because it decides what a realistic goal looks like.

Your score moves in scaled points, and roughly two extra raw marks is worth one scaled point, from the scoring data on our TARA score conversion page.

So a two-mark improvement between mocks is inside the measurement noise. A five or six mark improvement is worth two scaled points or more, and that is the size of movement to judge yourself on.

All of it, honestly, in one place. This is a short list, and the shortness is the whole problem.

The two official TARA tests

UAT-UK publish two complete tests, free, hosted on Pearson VUE and running on the same software as the live exam.

A specimen test and a sample exam. Each covers all three modules, so you will sometimes see them counted as six papers, but two full sittings is what they really are.

They are the most accurate practice that exists, because they run on the real thing. There are also only two of them, and once you have sat one under timed conditions it is spent.

You cannot un-see a question.

The free TSA and BMAT archive

UAT-UK also host the two tests TARA was built from, free and ungated.

Here's how that breaks down:

That is 37 free papers, which is more material than most people get through, and it is a foundation to build on rather than a finished plan.

The catch is that a mark on a 2009 BMAT paper is measuring something next door to TARA rather than TARA itself.

Every paper, with answer keys and the years ranked by how closely they still match, is on our TARA past papers page.

Our comparison of TARA, TSA and BMAT goes through the question types line by line if you want the detail behind the table above.

"TARA is hard" and "TARA is fine" are both useless as preparation advice. The useful question is which parts are hard, and some of that is the same for everybody.

Parallel Reasoning is never easy. Not once, in either official paper.

The format carries a floor, because you have five mini-arguments to formalise before you can start comparing anything.

Principle is never easy either. Pulling an unstated general rule out of an argument and recognising it wearing different clothes is hard every time it appears.

Assumption never gets truly hard in the official papers, though nothing says it could not.

Problem Solving does the same thing with shapes. Every 3D and geometry question in both official papers sits at the top of the difficulty range.

There is no gentle spatial question on the real test. If that is the part you dread, the good news is that it is genuinely hard rather than you being slow at it.

Two things follow for how you spend your weeks.

Front-load the types that are hard for everyone, because they will not get easier by being left until week ten. Our Critical Thinking guide covers all seven types and the Problem Solving guide covers the three skills.

Then find the ones that are hard for you specifically, which is a different list for everybody and is the entire reason to track your practice by question type rather than by score.

One thing that is not a difficulty factor, and it is worth saying plainly.

Nothing on either multiple-choice module requires a fact you have to know going in. The official papers draw their topics from general interest material: transport policy, charities, sport, health, the gig economy.

Reading widely is a pleasant thing to do. It is not TARA preparation.

Twelve weeks is not us being cautious. It is the shortest honest answer to what this test asks of you.

TARA rewards a way of thinking that sits slightly outside how school trains you: reading an argument for what it actually commits to rather than what it seems to say, and finding the route through a problem rather than applying a method you were taught.

That shifts slowly. It is a habit rather than a topic, and habits take weeks of short, regular contact to move, which is precisely why cramming does so little here.

A fortnight of panic will teach you the format and almost nothing else.

The plan also exists to protect one thing: the two official tests, spent last and spent properly.

Here is the shape before the detail:

1. Weeks 1 to 3: learn the types, untimed

Do not put a clock on anything yet. Speed built on top of the wrong method is just a faster way to be wrong.

Work through the seven Critical Thinking types and the three Problem Solving skills one at a time, in sets of about 20 to 25 questions, with the solutions read properly afterwards.

Twenty questions understood beats sixty marked and forgotten.

Use TSA Section 1 papers for the reasoning work and BMAT 2020 to 2023 for problem solving, since those four are the closest format match in the archive.

A dedicated TARA question bank does one thing the archive cannot do here, and this is the block where it matters most.

Archive papers are not tagged by TARA question type. You can mark a TSA paper and still not know whether you are losing marks on Assumption or on Parallel Reasoning, which is exactly the thing this block is supposed to tell you.

Practice written to the TARA format is categorised by type, so your weak spots show up by name instead of as a score. Ours is free to start if you want to work that way: create an account and pick a type.

By the end of week three you should be able to name the question type within a few seconds of reading it. That is the goal for this block, not accuracy.

2. Weeks 4 to 8: drill against the clock

This is the longest block and it is where most of the improvement happens.

Switch to timed sets from week four. Short and frequent beats long and occasional, because the thing you are training is recognition speed, and that responds to repetition rather than to endurance.

Three sessions a week gets you through a meaningful volume comfortably. Two a week works if you started early enough.

Split your time towards your weaker module. You will know which one it is by now, and the scores are reported separately, so there is no total to hide a weakness inside.

Sit something to time every week from week four, even if it is only one 40-minute module. Untimed accuracy tells you nothing about what happens at minute 38.

3. Weeks 9 to 10: full papers, real conditions

Now put the modules together, because the real test runs 120 minutes in a fixed order and fatigue in the third module is part of what is being measured.

Real conditions means real conditions: no pauses, no looking anything up, no phone, and nothing to write on but a whiteboard.

That last one catches people out. The test is on screen at a Pearson VUE centre, so you cannot annotate a passage or scribble beside a diagram the way you can on paper.

Our practice test guide covers how to run a single mock properly, and it is worth reading once before the first one.

4. Weeks 11 to 12: spend the official material

Only now do you open the two official tests.

Sit the specimen test as one full sitting, all three modules in order, to time. Review it properly over the following few days.

Then sit the sample exam the same way, about a week later.

Those two sittings are the only honest readiness check you get, because they are the only genuinely TARA material in existence and they run on the real software.

Do not spend them in week one as a diagnostic. A diagnostic tells you something you will know anyway by week three, and it costs you the one thing you cannot get back.

If you reach the end with time still on the clock, spend it on more questions. Volume is most of what turns a rattled sitting into a comfortable one, because every question whose shape you already recognise is one fewer surprise on the day.

A lot of people find this page late. The plan below is not as good as twelve weeks, and it is much better than panicking.

If you have about a month

Week 1. Read the question types for both modules and work a small untimed set of each.

Do not skip this to get to the papers faster, because it is what makes the papers teach you anything.

Weeks 2 and 3. Small timed sets every day rather than bigger ones every other day.

With a month on the clock the repetition is doing the work, and a short set daily builds recognition faster than the same total volume in three long sittings.

Weight them towards your weaker module, and sit one full archive paper at the end of each week under real conditions.

Week 4. The official specimen test, to time, all three modules in order.

Review it. Then the sample exam two or three days before you sit.

Still keep both official tests for this final week. The temptation to open them early gets stronger the less time you have, and it is still the wrong call.

If you have a week

Keep both modules in the plan. They are scored separately with no total between them, so abandoning one to rescue the other just guarantees a low number next to your name on that half of the result.

What a week changes is the pace, not the coverage. You are doing more in less time, so the sets get shorter and the review gets faster, and nothing gets dropped entirely.

Weight it towards your weaker module, which for most people is Problem Solving.

Students generally find it the harder of the two, and the first year's data agrees:

28% of candidates never genuinely attempted at least one Problem Solving question, against 4% on Critical Thinking.

Something like two thirds of your time on the weaker module and one third on the stronger works better than all of one and none of the other.

Do one timed 40-minute set a day, and review it the same day. The review is the part that does the work, and it is the first thing people cut when they are rushed.

Sit one official test to time, in order, two or three days before. Not the night before.

You want a day in between to fix anything it exposes, and you want to arrive rested rather than raw.

Read the pacing section below whatever else you skip. It is the highest return per minute of anything on this page.

This is the most actionable thing published about the TARA, and almost nobody uses it.

In the first live cycle, 28% of candidates either never reached a Problem Solving question or spent under five seconds on one.

Five seconds is not enough to read one of these questions, let alone work the three or four steps it takes to answer it. Those are guesses.

Here's how that breaks down:

Four per cent against twenty-eight. More than a quarter of candidates left at least one Problem Solving question never genuinely attempted, against one in twenty-five on Critical Thinking.

The median Problem Solving time was exactly 40:00. Half of all candidates used every second and were still going when the clock stopped.

Across both modules, more than 85% used between 35 and 40 minutes.

UAT-UK are not hiding from this. Their report says the test is expected to be "slightly speeded", meaning that for many candidates the limit on their score is how fast they work rather than what they can do.

The rest of this section is what we would actually do about it.

The pacing that follows from the numbers

An even split is 1 minute 49 seconds a question. That is 40 minutes across 22, and the data says the average candidate finishes with no slack at all.

Nobody works evenly, and you should not try to. Some questions land in twenty seconds and others deserve three minutes.

What matters is the running total, so use one checkpoint rather than watching the clock.

Aim to have answered 11 questions when 20 minutes have gone. At 8 you are behind and need to start cutting losses.

At 14 you have time in hand and can afford to slow down on something that deserves it.

Set that checkpoint before you start and check it once. Checking the counter every question is its own kind of time sink.

The realistic target for a routine question is about 90 seconds. Manage that on the eighteen or so you find straightforward and you free up five or six minutes across the module, which turns the four hardest from under two minutes each into a little over three.

That is the whole game. Not going faster overall, but going faster where it is cheap and spending the difference where it is not.

Practice the decision to leave a question

The most expensive mistake in a timed reasoning test is staying with a question because you have already invested in it.

Five minutes on question 6 does not make question 6 more likely to come good. It makes questions 20, 21 and 22 more likely to go badly, and they are worth exactly the same mark.

Here is the habit that fixes it. When you have read the same option three times, you are not making progress.

Eliminate whatever you can, pick the best of what is left, flag it, and leave. Flagging does not change your answer, so moving on costs you nothing and you keep the option of coming back.

Then, at the end of the module, work your flagged questions in order of how close you felt. The ones where you were choosing between two options are worth revisiting.

The ones where all five still looked possible mostly are not.

And never leave anything blank. A blind guess between five options is one in five.

Rule out two and it is one in three. With about two minutes left, stop reasoning and sweep everything still empty, then go back to thinking if there is time.

None of this works if you try it for the first time in the exam. A timed practice set rehearses the decision to leave a question, which is uncomfortable and does not get easier by being read about.

The five, before the detail:

1. Review the questions you got right for the wrong reason.

2. Log mistakes by type, not by question number.

3. Work the wrong options, not just the right one.

4. Keep sessions short and frequent.

5. Sit something to time every week from week four.

1. Review what you got right for the wrong reason

On a five-option multiple choice question, a lucky guess and a correct method look absolutely identical in your score.

They do not look identical in the exam, because the method repeats and the guess does not.

So mark by method rather than by letter. If you cannot say why the right answer is right, you did not get that question, whatever the mark says.

2. Log mistakes by type, not by question number

"I got 14 out of 22" tells you nothing you can act on. "I lost four marks on Parallel Reasoning and two to careless misreading" tells you what to do on Saturday.

Five careless misreadings and five genuine reasoning gaps need opposite responses. One is a pacing problem and the other is a method problem, and treating them the same way fixes neither.

Track by question type from the start. It also tells you which module is weaker, which is the single most useful thing to know early, because the two scores are reported separately.

3. Work the wrong options, not just the right one

The skill being tested is discrimination between five plausible answers. An answer key gives you a letter, which is the one thing you already had a one-in-five chance of guessing.

Every wrong option on the real papers is built to look right to somebody who made a specific, predictable mistake.

Knowing which mistake you made matters more than knowing you got it wrong. So for each wrong option, name what it was designed to catch, in a sentence.

4. Short and frequent, not long and occasional

The thing being trained is recognition speed, and recognition responds to repetition rather than to endurance.

Twenty-five questions with the solutions read properly beats sixty marked and forgotten, every time.

It is also why cramming does not work here. Habits of reading take a few weeks to shift no matter how clever you are, and a weekend of panic shifts nothing.

5. Sit something to time every week from week four

Untimed accuracy tells you nothing about what happens at minute 38, which is where the 28% figure comes from.

One 40-minute module a week is enough. It keeps the pacing checkpoint honest and rehearses the decision to abandon a question, which is the habit that does not survive being read about.

Come back to the questions you got wrong a week later, without looking at the solution. If you can now say why the right answer is right and why your original pick was designed to catch you, it has stuck.

If you pick the same wrong option again, it has not.

It is not marked, so it gets ignored. That is the mistake.

Your response is sent as text to every university you applied to, and admissions tutors read it. Not scoring something is not the same as not using it.

Only two of the sixteen TARA courses have said publicly whether they use it, which means for fourteen of them you are guessing. Our guide to whether the Writing Task actually matters for your course goes through all sixteen.

The good news is that preparing for it is cheap.

Write to time, at least a few times. Forty minutes, up to 750 words, treating the limit as a real constraint rather than a soft target.

Spend the first five to ten minutes planning, before you type a word. That feels like giving away a quarter of your time and it buys back more than it costs, because the thing being read is the structure of your argument.

Type it, do not handwrite it. The real task is typed on screen at a test centre, and your writing speed on a keyboard is part of what the 40 minutes has to cover.

The structure every prompt follows, and a paragraph-by-paragraph skeleton, are in our TARA Writing Task guide. That page owns the technique, so this one stops here.

One practical note: there is no archive analogue for the Writing Task. The TSA and BMAT essays were a different task with a different brief, so the only genuinely relevant material is the two official prompts and whatever practice prompts you can find.

Preparation time is finite, so here is what we would cut first.

Redesigning your revision around a two-mark jump between mocks. Every mark counts on the day, and a single scaled point is a real difference to an admissions team sorting hundreds of similar applicants.

So chase every mark in the exam itself.

But a two-mark move between two practice papers sits inside the measurement error, so it is not yet evidence that anything has changed.

Judge your progress on five or six mark movements, which are worth two scaled points or more and are the kind that survive the error bar.

Any score predictor or conversion table. The real conversion is built after the window closes, for the specific version you sat, and is never published.

A tool that hands you one confident decimal is hiding a guess. The honest version, in three difficulty scenarios, is on our score conversion page.

Grinding questions without reviewing them. The review is where the learning is.

Doing 200 questions in a fortnight and marking them by letter teaches you almost nothing, and it burns material you cannot get back.

Preparing only for your stronger module. The scores are reported separately and there is no total, so a 7.5 cannot carry a 3.5.

It is the most natural mistake to make and the most expensive.

Pre-2020 BMAT papers as a format rehearsal. They are fine for volume and they are free, so use them if you want more questions.

Just do not treat a 2009 paper as a dress rehearsal for a test with a different shape.

Reading widely to "prepare". Neither multiple-choice module requires a fact you have to know going in.

Read widely because it is worth doing, not because you think it is on the test.

Opening the official tests early. There are two of them and you only get to sit each one blind once, so it is the one genuinely irreversible mistake available to you, and the easiest one to make in week one.

Only if live teaching is how you actually learn, and that is a question about you rather than about the test.

TARA does not require subject knowledge, so a tutor is not teaching you content you are missing. What they are giving you is real-time interaction and attention on your specific gaps, which for some people is genuinely how the penny drops.

For most people, most of the time, the binding constraint on a TARA score is volume of reviewed practice under a clock, and that is something you can do on your own with the free archive and a timer.

If you are deciding what to build that practice on, we compare every TARA option, ourselves included, in the best TARA preparation in 2026.

Whatever you pick, the plan on this page is the same. The method does not change with the material.