Admissions tests

What Is a Good TARA Score? Oxford and UCL Benchmarks (2027 Entry)

Published: August 18, 2026
Updated: September 1, 2026
By Miguel

The TARA has run twice, for one university, and almost everything written about what counts as a good score predates the only data that exists. UAT-UK published its first TARA technical report in August 2026, and it contains the real numbers: means, quartiles and deciles for both scored modules across the October 2025 and January 2026 sittings. This guide starts there, then does the thing nobody else has done: it converts Oxford’s historical Thinking Skills Assessment benchmarks, from the test the TARA replaces, into TARA targets for each of the ten Oxford courses that now require it.

To find out where you currently stand, work through the official TARA past papers and resources, sit a timed paper on our TARA practice tests, or start with our guide to how to prepare for the TARA.

How the TARA is scored

The TARA is three separately timed 40-minute modules taken in a single 120-minute sitting. Critical Thinking is 22 multiple-choice questions, Problem Solving is another 22, and the Writing Task is one essay chosen from three prompts with a 750-word limit. Every question is worth one mark and nothing is deducted for a wrong answer, so the maximum raw mark is 22 per scored module. You cannot return to a module once its clock runs out, and no calculator or dictionary is allowed.

Three modules, 40 minutes each
22 questions per scored module
Two separate scores, 1.0–9.0
Writing Task unscored by UAT-UK
There is no overall TARA score

Two things trip people up. First, Critical Thinking and Problem Solving are reported as two independent scores on the familiar UAT-UK scale from 1.0 to 9.0, to one decimal place. There is no combined or overall TARA score, so any “overall TARA score” you see quoted has been invented. Second, UAT-UK does not mark the Writing Task at all. Your response is passed to each university you have applied to, unmarked, to use as it sees fit.

The scale itself is built rather than observed. UAT-UK fixes the median of the October cohort at 4.5 and the 90th percentile at 7.0, then applies the same linear transformation of the underlying ability estimate to every later sitting in the same cycle. That is why 7.0 is the number that matters: it is the designed top decile, not an accident of one year’s paper.

The first official TARA percentiles

UAT-UK’s technical report for the 2025/26 cycle gives the numbers for the first time. A total of 2,626 candidates sat the TARA across the two sittings, 976 in October 2025 and 1,650 in January 2026, making it comfortably the smallest of the three UAT-UK tests: the ESAT drew 14,193 candidates and the TMUA 16,092 in the same cycle.

Module and sitting	Mean	Median	Top 25% from	Top 10% from
Critical Thinking, October 2025	4.46	4.5	5.8	7.0
Critical Thinking, January 2026	4.06	4.0	5.2	6.6
Problem Solving, October 2025	4.66	4.5	5.8	7.0
Problem Solving, January 2026	4.76	4.8	5.7	7.0

Quartiles and deciles from UAT-UK’s TARA technical report for the 2025/26 cycle, the first published TARA distribution data.

The bottom quartile is the figure nobody quotes: in October 2025 a quarter of candidates scored 2.9 or below on Critical Thinking and 3.5 or below on Problem Solving. Across the full cycle the median settled at 4.0 for Critical Thinking and 4.7 for Problem Solving. The January cohort scored materially lower on Critical Thinking, and because UAT-UK deliberately reused October’s scaling constants rather than re-anchoring, that gap is a real difference between the two cohorts rather than a change in the paper.

One further figure from the technical report deserves more attention than it gets. The standard error of measurement on a scaled score is roughly 0.9 to 1.2 points. On a nine-point scale that is large. If you score 5.5, your underlying ability sits somewhere between about 4.5 and 6.5 roughly two thirds of the time. Half-point differences between two candidates, or between two of your own practice sittings, are noise.

What counts as a good TARA score?

TARA score	What it means
Below 3.0	Bottom quartile of the first TARA cohort on Critical Thinking. Well below every Oxford benchmark we can reconstruct.
3.0 – 4.4	Below the designed median. Around the level of the average January 2026 candidate on Critical Thinking.
4.5	Exactly the designed median of the October cohort. Not a bad score, but not yet a competitive one.
5.0 – 5.8	Above average. By 5.8 you are into the top quarter of the field.
6.0 – 6.9	Strong, and around the level we estimate for the less competitive Oxford TARA courses and for UCL Computer Science.
7.0	The designed 90th percentile, and the direct heir of a TSA 70. This is the number to aim at.
7.5 – 8.5	Above almost every historical Oxford offer-holder benchmark once converted onto the TARA scale.
9.0	The ceiling. Scores are capped here.

The honest headline is that 7.0 is the number to aim at and 6.0 is the number to be comfortable with. Anything below the mid-fives puts you behind the historical offer-holder pack on every Oxford course we can reconstruct. And because there is no overall score, both modules have to clear the bar: a 7.5 on Problem Solving does not rescue a 4.0 on Critical Thinking, because tutors see two numbers rather than an average.

Who requires the TARA, and why the cohort is about to change

For 2026 entry the TARA was used by exactly one university: UCL, for Computer Science BSc and MEng, Computer Science and Mathematics MEng, and Robotics and Artificial Intelligence MEng. That matters enormously for how you read the numbers above. On the figures UCL publishes, the UCL Computer Science acceptance rate is 5.7%, or 4,006 applications for 228 places. The 976 candidates who sat in October 2025 were, in the main, UCL Computer Science applicants, which is precisely why that cohort scored higher on Problem Solving (4.66) than on Critical Thinking (4.46).

University	Courses requiring the TARA for 2027 entry
Oxford	Economics and Management Experimental Psychology History and Economics History and Politics Human Sciences Philosophy and Linguistics Philosophy, Politics and Economics PPE with Foundation Year Psychology and Linguistics Psychology and Philosophy
UCL	Computer Science BSc and MEng Computer Science and Mathematics MEng Mechanical Engineering BEng and MEng Robotics and Artificial Intelligence MEng

From the October 2026 sitting Oxford joins for ten courses and UCL adds Mechanical Engineering. Oxford’s TARA courses alone drew roughly 1,885 PPE applicants, 1,421 for Economics and Management, 752 for Experimental Psychology and PPL and 164 for Human Sciences in recent cycles; our Oxford and Cambridge admissions statistics has the full application and offer counts for each of them. The cohort is about to grow four- or fivefold and to tilt sharply towards humanities and social science candidates, so the Critical Thinking distribution will move. Because UAT-UK re-anchors the scale to each year’s October cohort, October 2026’s 4.5 will not be October 2025’s 4.5. No other university has adopted the TARA, and Cambridge has not.

Sitting the TARA?

Knowing the benchmark is not the same as reaching it. Our tutors sat the TARA themselves and coach it one to one, against the criteria the universities actually apply.

Book a free consultation

The TSA proxy: what Oxford’s offer holders used to score

Oxford has no TARA outcome data and will not have any until spring 2027 at the earliest. What it does have is more than a decade of Thinking Skills Assessment data, and the TSA is the test the TARA directly replaces, right down to the two modules and the unmarked essay. The two scales are built the same way, which makes the translation more than a guess.

Benchmark	TSA (0–99)	TARA (1.0–9.0)
Typical candidate for a highly selective course	around 60	4.5
Comparatively high	70	7.0
Very exceptional	80+	around 8.5+

Cambridge Assessment’s own explanation of TSA results said that typical applicants to the most selective UK courses scored around 60, that 70 was “a comparatively high score”, and that “only a few very exceptional applicants” passed 80. UAT-UK’s TARA scale fixes the median at 4.5 and the 90th percentile at 7.0. Line those anchors up and a TSA 70 and a TARA 7.0 are making the same claim about a candidate. The interpolation works out at roughly one TARA point for every four TSA points.

Economics and Management

For 2024 entry, Economics and Management applicants averaged 58.3 overall on the TSA, shortlisted candidates 67.7 and offer holders 69.8. That offer-holder average sits almost exactly on the “comparatively high” mark, which converts to a TARA 7.0. E&M rates the admissions test “High” in its shortlisting weighting, above GCSE profile and predicted grades, and it is the most competitive TARA course at Oxford: 1,421 applicants, 280 shortlisted and 97 offers for 2026 entry, an offer rate of under 7%.

Human Sciences, and the rescaling problem

Human Sciences publishes the cleanest weighting statement of any Oxford TARA course. For candidates with a contextualised GCSE score, the test and the GCSE score are given equal weight in the pre-interview score; a threshold is then set by the number of places available, and everyone above it is shortlisted. Where a candidate has no contextualised GCSE score the test is double weighted. That is an admissions test operating as a roughly 50% mechanical gate, and it is the single clearest reason to take the TARA seriously.

Human Sciences also illustrates why the raw TSA series cannot be read as a trend. Its offer holders averaged 73.9 in 2022, 61.5 in 2023, 63.5 in 2024 and 55.1 in 2025, and Economics and Management shows the same collapse between 2025 and 2026 entry. Oxford changed its TSA reporting basis at least twice across those cycles, including one year in which the two sections were reported out of 50 rather than out of 100. The drop is almost certainly a rescaling artefact rather than a decline in candidate quality, and Oxford publishes no explanation. Anyone quoting a five-year TSA trend line is quoting a formatting change.

Experimental Psychology and PPL

The joint Experimental Psychology and PPL reports are the only Oxford admissions-test data published with interquartile ranges, which makes them the most useful rows in this whole exercise. For 2025 entry, applicants averaged 57 with a middle half of 50 to 64; shortlisted candidates averaged 66 with a middle half of 62 to 70; offer holders averaged 67 with a middle half of 62 to 72.

The only Oxford admissions-test data published with interquartile ranges. The offer-holder middle half sits entirely inside the applicant range: a strong score moves you a long way and guarantees nothing.

Oxford course (TSA overall)	All applicants	Shortlisted	Offer holders
Economics and Management, 2024 entry	58.3	67.7	69.8
Human Sciences, 2022 entry	61.0	65.9	73.9
Human Sciences, 2024 entry	58.0	63.0	63.5
Experimental Psychology and PPL, 2023 entry	59	65	67
Experimental Psychology and PPL, 2025 entry	57	66	67

Three of Oxford’s TARA courses have never had their scores published by stage at all: History and Economics, History and Politics, and PPE. PPE publishes applicant averages split by whether the candidate took A level Maths, but the applicant, shortlisted and offer-holder comparison appears only as unlabelled graphs. What PPE does state is that the pre-interview test is weighted “High” in shortlisting and that applicants who do not sit it “will not normally be shortlisted”. In the 2023-24 cycle, no applicant without a test score was interviewed or made an offer.

Our TARA targets for 2027 entry

Converting the TSA benchmarks above onto the TARA scale, then adjusting for how competitive each course is and for the fact that the October 2026 cohort will be dominated by exactly these applicants, here is where we would tell a student to aim. These are Clavis estimates, not published thresholds. Neither Oxford nor UCL has published a TARA threshold, an expected score, or any guidance beyond the standard line that no score guarantees an interview.

Applying for	Realistic target	Stretch target
Oxford Economics and Management	7.0	7.5+
Oxford PPE (and PPE with Foundation Year)	6.5	7.0+
Oxford Experimental Psychology and PPL	6.5	7.0+
Oxford Psychology and Philosophy, Psychology and Linguistics, Philosophy and Linguistics	6.5	7.0+
Oxford History and Economics	6.5	7.0+
Oxford History and Politics	6.0	6.5+
Oxford Human Sciences	6.0	6.5+
UCL Computer Science (BSc, MEng, with Mathematics)	6.0	6.5+
UCL Robotics and Artificial Intelligence	5.5	6.0+
UCL Mechanical Engineering	5.5	6.0+

Aim for both modules, and treat Problem Solving as a speed test. The technical report shows that 8% of candidates left at least one Problem Solving question unreached, against 1% on Critical Thinking, and that the median candidate used all 40 minutes. Critical Thinking is where preparation transfers most cleanly, because its seven question types are fixed and published: identifying the main conclusion, drawing a conclusion, identifying an assumption, assessing additional evidence, detecting reasoning errors, matching arguments and applying principles.

Do not switch off for the Writing Task. UAT-UK does not mark it, but Oxford has marked admissions essays in-house for years, including TSA essays for PPE and LNAT essays for Law, and UCL states that it uses all three modules. Write it properly.

Find this article useful? Check out our other resources for students sitting the TARA.

TARA Practice Tests

Practise the TARA under timed conditions with original questions.

TARA Past Papers and Resources

Official TARA materials plus the archived TSA and BMAT papers UAT-UK recommends.

How to Prepare for the TARA

Format, dates, question types and a full preparation plan for the TARA.

Sources

Frequently asked questions

What is a good TARA score?

7.0 is the benchmark, because UAT-UK builds the scale so that 7.0 is the 90th percentile of the October cohort. In the October 2025 sitting the median was 4.5 on both scored modules and the top quarter began at 5.8. Anything from 6.0 upwards is a strong score; anything below the mid-fives is behind the historical Oxford offer-holder pack.

What is a good TARA score for Oxford?

Oxford has published no TARA data and no threshold, because 2027 is its first cycle using the test. Converting its historical TSA benchmarks onto the TARA scale, we would target 7.0 for Economics and Management, 6.5 for PPE, Experimental Psychology, PPL and the psychology and philosophy joint courses, and 6.0 for Human Sciences and History and Politics. Those are our estimates rather than Oxford figures.

Is 4.5 a good TARA score?

4.5 is exactly the designed median, so it is an average score rather than a good one. It is not disqualifying: Oxford states that no test score guarantees or rules out an interview, and its offer-holder ranges have always overlapped heavily with the applicant field. But on every Oxford course we can reconstruct, the average offer holder sat well above the median of the field.

Is there an overall TARA score?

No. Critical Thinking and Problem Solving are reported as two separate scores, each from 1.0 to 9.0, and UAT-UK publishes no composite. Any source quoting a single overall TARA score has made it up. This matters practically: a very strong score on one module will not offset a weak score on the other, because admissions tutors see both.

Is there a minimum TARA score for Oxford or UCL?

Neither university publishes one. UCL says the TARA “will be considered alongside your A-Level grades and UCAS Personal Statement as part of a holistic admissions review” and states no cut-off. Oxford’s standing line on its predecessor test was that “it is not the case that achieving a certain score will guarantee an interview”. In practice, courses such as Human Sciences do use the test as a ranked threshold for shortlisting.

How many raw marks do I need for a 7.0?

Nobody outside UAT-UK knows. There are 22 marks available in each scored module, but UAT-UK publishes no raw-mark-to-scaled-score conversion table for the TARA, and the scaling is a Rasch transformation that varies with question difficulty rather than a fixed percentage. Any site quoting a specific raw mark for a 7.0 is extrapolating from the old TSA.

Does the TARA Writing Task count?

UAT-UK does not mark it and it produces no score. Your response is sent unmarked to each university you have applied to. Oxford has a long track record of marking admissions essays in-house, including TSA essays for PPE and LNAT essays for Law, and UCL states that it uses all three modules. Write it properly.

When do I sit the TARA for 2027 entry?

The main sitting is 12 to 16 October 2026, with booking closing at 6pm UK time on 28 September 2026 and results released on 16 November 2026. A January 2027 sitting runs from 4 to 8 January for Oxford Foundation Year and some mature college applicants. The test costs £78 in the UK and Ireland and £133 elsewhere, with bursaries available.