Why Oxford Replaced the MAT with the TMUA

This is the story and analysis of the switch; for the side-by-side format comparison see TMUA vs MAT, and for Oxford's current requirements and target scores see TMUA for Oxford. The January 2026 announcement that Oxford was retiring the MAT ended one of the longest-running fixtures in UK maths admissions, and it left a lot of applicants asking the same two questions: why did this happen, and what does it change for me? This guide answers both, separating what is officially confirmed from what is structural analysis, and finishing with what an applicant should actually do about it.

What happened, and when

The confirmed facts are these. The MAT (Mathematics Admissions Test) was Oxford's own admissions paper for Mathematics and Computer Science, running from 2007 to 2025, nearly two decades. In January 2026, Oxford announced that the MAT is discontinued, and that from the October 2026 sitting its applicants for Maths and Computer Science, including the joint courses, take the TMUA instead. The first affected cohort is everyone applying for 2027 entry.

Here is the timeline in one view:

The scale of the change is easy to understate. The MAT was not a minor assessment; it screened two of the most competitive courses in the country for nineteen years and built a deep archive of papers, solutions and folklore. All of that is now the history of a test nobody sits.

Why did Oxford switch? The structural logic

Oxford's announcement was not accompanied by a long public rationale, so honesty requires a distinction: the fact of the switch is official; the reasons below are analysis of the structural forces that are plainly visible, not quoted Oxford policy.

The sector had already consolidated around UAT-UK. By the time of the announcement, admissions testing for mathematical subjects had reorganised around a single shared vehicle: UAT-UK, the joint venture between Cambridge and Imperial that took over the TMUA in 2024. Cambridge, Imperial, LSE, Warwick, Durham and UCL were already using the TMUA for maths, computer science and economics courses. That left Oxford as the last university running its own bespoke maths test for exactly the cohort a shared national test already served. Joining the shared test made Oxford the seventh user rather than the lone holdout.

On-screen delivery at scale is the direction of travel. The TMUA is sat on screen at Pearson VUE test centres, a network built to deliver standardised tests to large numbers of candidates worldwide with consistent conditions and security. Running a single-university test means owning question production, delivery, marking and results-handling for one applicant pool every year. Handing that machinery to a shared national provider is the same consolidation logic that produced UAT-UK in the first place.

One test is better for applicants, and Oxford competes for the same applicants. Before the switch, a student applying to Oxford and Cambridge for maths sat two different admissions tests in the same autumn, with two preparation tracks. Now one October TMUA covers Oxford, Cambridge, Imperial, Warwick and the rest in a single sitting. A shared test lowers the burden on exactly the strong, heavily-oversubscribed cohort every one of these universities is trying to attract.

None of this required the MAT to be a bad test. It was, by broad consent, a well-liked exam that trained a generation of applicants in sustained problem-solving. It was retired because a lone bespoke paper stopped making structural sense, not because the mathematics it tested stopped mattering.

What the switch means for applicants

The practical consequences are bigger than the announcement made them look.

Oxford and Cambridge maths applicants now sit the same test. For the first time, one October score serves both applications (and Imperial, Warwick and Durham besides). That is genuine convenience, but it cuts both ways: there is no longer an Oxford-specific test you can play to, and your one TMUA performance is read by every TMUA university on your form.

October is the only window that counts. Oxford, like Cambridge, does not accept the January sitting for standard applicants (it exists for mature and Foundation-Year candidates), so the 12-16 October 2026 window is your single shot in the cycle, arriving earlier in the autumn than many MAT-era applicants are used to planning around.

The field sitting the TMUA is larger and stronger. Oxford Maths and Computer Science applicants, one of the strongest cohorts in the country, have been added to a test already sat by Cambridge, Imperial, LSE, Warwick, Durham and UCL applicants. Whatever else follows, the company you are keeping on results day has improved.

The MAT archive is no longer the right preparation. Two decades of MAT papers remain online, and they are now a trap for the unwary. The MAT was a long written paper with method marks; the TMUA is two 75-minute all-multiple-choice papers with no marks for working, plus an entire paper (Paper 2) devoted to logic and proof that the MAT never isolated. MAT problems can still exercise the underlying maths, but as exam preparation they train the wrong format, the wrong pacing and the wrong skills. The full reckoning of what transfers is in TMUA vs MAT.

Here is the sharpest example of what changed: a genuine TMUA Paper 2 question of the kind the MAT never asked. This is the paper MAT-era preparation leaves completely untouched, so try it before revealing the solution:

What it means for the TMUA itself

The honest answer is that the TMUA has become more important and more competitive at the same time. It is now the single gateway test for maths and computer science at both Oxford and Cambridge, plus Imperial and the rest, which concentrates more admissions weight on one October sitting than any UK maths test has carried before.

On scores, some perspective helps. The TMUA reports each paper on a 1.0 to 9.0 scale and the national average has sat around 4.5. How the scale behaves as the cohort grows and strengthens is not something anyone can promise you, and nothing official has been published about it, so treat confident predictions about boundary shifts with suspicion. The practical response does not depend on that uncertainty anyway: in a stronger field, you prepare against the hardest genuine material rather than the average, and you make the trainable parts of the test, above all Paper 2 reasoning, into banked marks rather than question marks.

What to actually do

If the switch affects you, the plan is concrete.

1. Book the October sitting now. Oxford (like Cambridge) uses the October window: 12-16 October 2026 for 2027 entry. Booking opened on 20 July 2026 and closes on 28 September 2026 at 6pm BST, with the fee at £78 for candidates in the UK and Republic of Ireland and £133 internationally. The deadline lands mid-UCAS-season and is easy to miss. Full logistics in our dates and registration guide.
2. Learn the TMUA format before anything else. Two 75-minute papers, 20 multiple-choice questions each, no calculator, no method marks. Sit one official paper under time simply to feel the rhythm; the maths will be familiar, the format will not.
3. Make Paper 2 your priority. Logic, necessary and sufficient conditions, counterexamples and proof-error analysis are the biggest delta from anything the MAT era prepared people for, and they are the most trainable marks on the test.
4. Retire the MAT archive. Use official TMUA past papers, the specimen plus 2016 to 2023, as your calibration standard, and save some for full timed mocks near the exam.

For the work itself, CrackTMUA gives you 640+ practice questions, every official TMUA question plus 280+ original trap-based ones, each with a worked solution that names the trap and shows the fastest method, alongside 24+ full timed mocks in a replica of the real Pearson VUE screen. The free tier gives you 10 questions a day; Premium unlocks everything for £49 one-time, with 12 months of access. The MAT era is over; the applicants who adapt to the new test earliest will be the ones it rewards. Start practising free.

Frequently asked questions