How TMUA Scoring Works: From Rasch Scaling to University Data

Your TMUA results statement shows a single number: an overall score between 1.0 and 9.0, reported to one decimal place. Paper 1 and Paper 2 do not receive separate scores. UAT-UK equates and scales the two papers together to produce one overall score, a deliberate design choice and a clear point of contrast with the ESAT and TARA, where each module is equated and scaled separately.

Two practical consequences follow. First, there is no per-paper threshold. Universities never see a Paper 1 score or a Paper 2 score, so a rough 75 minutes on Mathematical Reasoning cannot 'fail' you on its own; what counts is your performance across all forty questions taken together. Second, every question across the two papers carries equal weight and there is no penalty for wrong answers, so the official advice is simply to attempt everything.

The scale is capped at both ends: the lowest reportable score is 1.0 and the highest is 9.0, as set out in the official Explanation of Results document. One misconception is worth clearing early. The 1 to 9 score is not a percentage of questions answered correctly, nor a rebadged raw mark. It is a statistically scaled score, and the next section explains what that scaling actually does and why it exists.

The official description of scoring has two steps. Step one is refreshingly blunt: your final score is based on the number of correct answers you give, nothing else. There is no negative marking and questions across both papers carry equal weight, which is why UAT-UK advises candidates to attempt every question, however unsure they feel.

Step two is where the statistics come in. Raw marks are not converted directly into the 1 to 9 scale. UAT-UK analyses the response data using the Rasch model, an item response theory model widely used in high-stakes assessment. No two test versions can be written to exactly the same difficulty, and no two sittings attract exactly the same field of candidates; the Rasch model quantifies those differences, and a process called equating then removes them. The result is that scores from different test versions and different sittings sit on one unified scale: a 6.0 from October and a 6.0 from January represent the same standard, which is what allows admissions teams to compare applicants across sittings fairly.

The practical takeaway for your preparation: UAT-UK publishes only the final scaled scores and the distribution, with no grade boundaries of any kind; and since equating adjusts for the actual difficulty of each test version, there is no fixed conversion between a practice-paper percentage and a scaled score, so projecting one from the other is guesswork. Use practice results to diagnose gaps and pacing problems, not to predict the number on your results statement.

Every sitting comes with an official document called the Explanation of Results, published by UAT-UK, and it is the single most authoritative guide to reading your number. The October 2025 edition states the design intent plainly: the scale is built so that typical candidates score around 4.5, roughly 10% of candidates score higher than 7.0, and scores are capped at 1.0 at the bottom and 9.0 at the top. The document also includes a histogram of the full cohort in half-point bands, so you can place your own result precisely.

Honesty requires flagging an error on the official website here. On the test-results page, one paragraph fixes the median at 4.5 and the 90th percentile at 7.0, while a later paragraph on the same page puts the 90th percentile at 9.0. Both cannot be true, and the formally published Explanation of Results settles it: approximately 10% of candidates score above 7.0, and the 9.0 figure is a typo.

One further piece of official framing matters for morale. UAT-UK says outright that its tests are designed to be challenging in order to separate highly capable applicants, and that candidates used to top grades in school exams should not expect to score as highly here. Read through that lens, a 4.5 is not a disappointing 'half marks'; it is the typical result, and anything above 7.0 is a top-decile performance.

UAT-UK closes every door that candidates instinctively look for. There is no pass or fail score for the TMUA. The organisation provides no grade boundaries of any kind, and scores are used by each university alongside the other information in your application, on that university's own terms. Nor is there a review route: UAT-UK states that its scoring process is rigorously tested and quality-assured before, during and after every testing window, and that it therefore does not accept appeals against test scores. Add one more rule to the picture: you may sit the test only once within an admissions cycle.

Three practical conclusions follow. First, any claim online that a particular score 'guarantees' an outcome is not an official position, because no official threshold exists anywhere. Second, the number you produce on test day is final, with no re-mark to request and no second sitting in the same cycle to fall back on, so the preparation phase is where the outcome is actually decided. Third, the only score information with a named, checkable source comes from the universities themselves, and only a minority of them publish anything at all. The next two sections cover everything that is currently on the record.

In 2024 the TMUA's reported scale changed, and every historic score has to be read through that change. Durham University spells out the conversion on its entry-requirements page: a score of 6.5 from 2023 equates to a score of 5.0 from 2024 onwards, and a 4.5 from 2023 equates to somewhere between 3.2 and 3.5 on the current scale.

The consequence is blunt: target scores, admissions anecdotes and forum threads from 2023 or earlier do not transfer. The classic trap is the widely repeated line that '6.5 or above is competitive'. That figure comes from the old scale, and by Durham's own conversion corresponds to roughly 5.0 today; adopting it unconverted means quietly inflating your target by about a point and a half, which distorts both your preparation plan and your read on practice results.

The opposite mistake is just as easy to make. Not every demanding target is stale data: Imperial's mean of 7.6 for Mathematics offer-holders is a 2025-entry figure on the current scale, both real and current. So treat every number you encounter as unverified until you can answer three questions about it: which year was it earned, on which scale, and who published it. A number that survives all three checks belongs in your planning; one that fails any of them belongs in the bin.

Imperial College London is currently the only university that publishes department-level score and outcome data, through a dashboard titled 'Admissions tests scores 2025' embedded in its undergraduate admissions pages. For 2025 entry, the mean TMUA scores of offer-holders were:

(Home non-WP means Home applicants outside the widening participation category.)

Two things matter when reading the table. First, if you are applying from outside the UK you count as Overseas, and the Overseas column is the one to benchmark against: in every department, Overseas offer-holders averaged higher scores than Home applicants, while Overseas offer rates were generally lower. Second, set the 7.6 against the official distribution: only about 10% of candidates nationally score above 7.0, so the average Mathematics offer-holder at Imperial sits inside the national top decile.

Imperial's own caveats should always travel with the data. This is historic information; score distributions and offer outcomes vary from year to year; the figures should not be read as requirements for current or future cycles; a high score does not guarantee an offer, and a low one does not sink an application. These are averages, not cut-offs.

Durham and Warwick publish no dashboards, but both have put usable statements on the record.

Durham, where the TMUA is recommended rather than compulsory for its four mathematics courses (G100, G103, G111 and G114), states that a score of 5.0 or above makes you automatically eligible to be considered for a reduced offer: A*AA with an A* in Maths or Further Maths, one grade below the standard A*A*A with A*A* in Maths and Further Maths. Below 5.0, down to a threshold set each year somewhere between 3.2 and 3.5, Durham still treats the score as positive evidence of mathematical ability, with a good chance of the standard offer.

Warwick requires every undergraduate mathematics applicant to sit either the TMUA or STEP, and reports that in the last admissions cycle the majority of offers went to applicants scoring 5.0 and above, with some offers made below 5.0 after a holistic assessment of the overall profile.

Put these alongside Imperial and the picture sharpens. On one and the same scale, published figures range from Durham's 5.0 to Imperial Mathematics' 7.6. Neither is wrong and neither is stale; they simply describe two very differently contested destinations. Score targets only make sense when set per university and per course, never as a single 'what do I need on the TMUA' number.

You do not need to count weeks to know when results arrive, because UAT-UK publishes the dates outright. For 2027 entry, the October sitting's results are released on 16 November 2026 and the January sitting's on 8 February 2027, both through your UAT-UK account. It is worth knowing that the official website disagrees with itself on the interval: the TMUA information page says results come approximately six weeks after the test, while the results page says approximately four. Ignore both and remember the dates.

Delivery to universities is fully automatic. Scores are sent to the institutions in your UCAS application that require the test, matched course by course, and candidates cannot choose which institutions see their score. Combine that with the one-sitting-per-cycle rule and the October-versus-January decision comes into focus: it is a choice of date, not a trial run, and the single number you produce will appear, unedited, in front of every relevant admissions team.

Cambridge and Oxford applicants generally have no choice at all: with narrow exceptions (mature-college applicants at Cambridge with a January deadline, and Oxford's Astrophoria Foundation Year), they must sit in October. Thinking these constraints through before you book is far more useful than agonising after results day.

Here is the whole article folded into a working method.