TMUA Scoring Decoded: A 6.5 Is Twice as Hard as in 2023, and the Median Chinese Candidate Rivals the UK Top 10%
The 2024-2026 numbers behind the new TMUA scale: the October-fixed 7.0, how rare a 6.5 has become, who fills the top bands, October versus January, and how to adjust your preparation.
TMUA moved to computers in 2024 and the scoring rules changed with it. Plenty of candidates still estimate their score the old way and find themselves well short on results day. These are the numbers that decide where you really stand.
1. 7.0 Means the October Top 10%
After each cycle, every candidate’s ability is converted to 1.0-9.0 along one straight line. Two points fix that line: the October median becomes 4.5 and the October top-10% boundary becomes 7.0.
| Cycle | Conversion |
|---|---|
| 2024-25 | score = 4.4897 + 1.7998 × ability |
| 2025-26 | score = 4.6046 + 1.6369 × ability |
January candidates are converted with October’s line. The stronger the October field, the further the line shifts and the lower the score the same ability earns.
2. A 6.5 Is Worth More Than Twice What It Was
| Year | Share of candidates |
|---|---|
| 2019 | 28.0% (6.5 or more) |
| 2020 | 25.6% |
| 2021 | 27.6% |
| 2022 | 31.2% |
| 2023 | 31.9% |
| 2024-25 new format | 11.9% (6.3 or more) |
| 2025-26 new format | 12.9% (6.3 or more) |
Under the old papers roughly one candidate in three or four reached 6.5. Now, even with the bar dropped to 6.3, only one in eight does. Pre-2024 benchmarks will make you overestimate where you stand.
3. Chinese and UK Candidates Are a Full Band Apart
| 2024-25 candidates | Number | Mean | Median | Top 25% from | Top 10% from |
|---|---|---|---|---|---|
| Chinese nationals | 2,554 | 5.42 | 5.4 | 6.7 | 8.4 |
| Hong Kong | 296 | 5.06 | 5.0 | 6.3 | 7.6 |
| Singaporean | 316 | 4.78 | 4.7 | 5.8 | 6.9 |
| Indian | 779 | 3.63 | 3.5 | 4.7 | 5.7 |
| Living in the UK | 7,715 | 3.86 | 3.8 | 4.8 | 5.8 |
| All candidates | 13,855 | 4.20 | 4.0 | 5.2 | 6.5 |
The median Chinese candidate (5.4) sits only 0.4 below the top-10% boundary for UK residents (5.8). The Chinese top-25% boundary, 6.7, is already above the top 10% of all candidates.
The top bands are even more lopsided: at least six in ten candidates scoring 7.4 or more live outside the UK and EU, and nearly six in ten scoring 8.3 or more are Chinese nationals, who are 18% of all candidates.
4. Several Times More Candidates Score 8.8 or More
In 2019-2023, 0.2-0.7% of candidates scored 8.8 or more each year, which meant missing no more than about two questions. In the new format 3.1% (2024) and 3.5% (2025) of October candidates score 8.8-9.0: 280-340 people a year inside the top 0.2 of the scale. For Cambridge, Imperial or LSE, a high score alone no longer sets you apart.
5. October and January: Three Times the High Scorers
| Sitting | Candidates | Median | Top 10% from | 7.3 or more |
|---|---|---|---|---|
| October 2024 | 8,987 | 4.5 | 7.0 | 8.6% |
| January 2025 | 4,868 | 3.3 | 5.4 | 2.6% |
| October 2025 | 9,685 | 4.5 | 7.0 | 9.2% |
| January 2026 | 6,407 | 3.6 | 5.9 | 3.3% |
Both sittings use the same conversion, so the scores compare directly. The difference is the field: only about 3% of January candidates reach 7.3. The official line that about 10% score above 7.0 describes October only.
6. Time and Difficulty: Half the Room Works to the Last 23 Seconds
The test lasts 2:30:00 and the median time used is 2:29:37; nine in ten candidates use more than 2:15. In 2024-25, 23% of candidates (3,227 people) answered at least one question in 10 seconds or less or never reached it; in 2025-26 it was 16%. Of the 186 questions used in 2024-25, 21 were answered correctly by fewer than 20% of candidates and only one by more than 90%.
7. 2027 Entry Brings a New Wave of Applicants
Universities using TMUA for 2025 entry: Cambridge, Imperial, Durham, LSE and Warwick. Added for 2027 entry: Oxford Mathematics, Computer Science, Mathematics and Computer Science, Mathematics and Philosophy, Computer Science and Philosophy; Cambridge Mathematics; UCL Economics.
8. Scoring Rules: Only the Number Correct Counts
Every question carries equal weight and wrong answers cost nothing, so never leave one blank. On the same form, the same number correct always gives the same score, and a missed hard question costs exactly as much as a missed easy one. Two people with the same raw mark and different scores sat different forms.
A single sitting carries a measurement error of about ±1.3 (95% interval). A one-point swing between sittings is common, and the half point between 6.5 and 7.0 sits inside that error.
9. How to Adjust Your Preparation
- Turn your target score into a raw mark: use the MioMath TMUA score calculator, aim for the top of the range your course needs and add a one-point margin.
- Train at exam pace: both papers back to back for 150 minutes, on screen, no calculator. Bank every question you can solve first and skip anything that stalls you for more than three minutes; see the two-pass skipping strategy.
- Use two layers of material: the official 2016-2023 papers for the base, and the 2025 reconstructed sets for the current format and difficulty.
Sources
Ready to Achieve Your Target Score?
Start preparing with MioMath's realistic mock exam system, featuring past paper practice, timed simulations, and detailed solutions.
Start Practicing FreeTable of Contents
Related Guides
TMUA October 2026 Score Estimate: What Your Raw Marks Are Likely Worth
A simple lookup table from total raw marks to a likely TMUA score, based on the 2024 and 2025 papers and a check of this year's recalled questions.
TMUATMUA October 2026 Recap: What the Recall Data Actually Shows About Form Overlap, Repeats and Results Day
Every October 2026 form came from one shared international question bank, and overlap between forms was wildly uneven: about 35%, about 10%, and one form with 0% overlap. Cross-year repeats were real but small (3-7%). What that means for difficulty, circulating answer keys, and the wait for mid-November results.
TMUATMUA October 2026 Prediction: A Calculation-Heavy Paper 1, a Logic-Leaning Paper 2, and How to Use Your Last Days
Our prediction, read from the 2024 and 2025 global papers: Paper 1 will carry a heavy load of surds and algebra, and Paper 2 will lean on logic but is manageable once you are used to it. Plus what Oxford moving from the MAT to the TMUA changes, and what to practise before you sit.