TMUA past papers 2016–2023: the anatomy of 320 official questions

Every question from all 16 official papers (8 years × Paper 1 & Paper 2), tagged to the canonical Strand › Topic › Subtopic › Skill spine and rated on a D2–D5 decision-load scale. For students who want to know how these exams are put together.
© 2026 Dr Nicky Grant. All analysis, difficulty ratings, skill tagging and figures on this page are original work and protected by copyright. Reproducing, republishing or otherwise reusing this data, these findings or these charts, in whole or in part, on another site, in print, or elsewhere, requires prior written permission. Request permission · Unlock & app-pass terms

Where are the TMUA past papers?

The official TMUA past papers from 2016 to 2023, with answer keys, are free on the UAT-UK website. No papers from 2024 or 2025 have been released. This report analyses every question in those 16 papers, 320 in total, by topic, skill and difficulty, so you can see what the papers test and choose which to practise first.

More free TMUA resources: all resources · 50 practice questions · timed mock exam · Paper 2 notes on logic and proof · why there are no 2024 or 2025 papers

Want 1:1 help turning this into a study plan?

I tutor TMUA candidates one-to-one, turning this question-by-question analysis into a personalised revision plan, working through weak skills and timed papers together.

See TMUA tuitionBook a free consultation

EntryQ: the TMUA question bank built from this research

EntryQ is the question bank built on this report: about 2,500 questions organised by the same topics and skills, 20 timed mock papers and a Plan that picks your next questions from your results, now open to pass holders. £30 to the January 2027 sitting, or £100 to the October 2027 sitting until 31 December 2026.From 1 January 2027: £175 for a one-year pass.

See the passes

Primary tag = each question counted once under its primary subtopic. All tags = counts primary and secondary tags, so one question can appear under several subtopics. Each chart is badged with the rule it uses.

Key findings static summary

The headline numbers from the full analysis, stated plainly. Every figure below is computed from the frozen dataset of all 320 official TMUA questions (2016–2023, 16 papers), each individually worked through and assessed by Dr Nicky Grant. The interactive charts in the rest of the report let you explore each one.

Watch the 10-minute walkthrough

Prefer it explained? Dr Grant talks through the headline findings, how difficulty has drifted since 2016, which skills dominate the papers, and the revision advice that follows from them. Then dig into the interactive charts below.
Video: TMUA Past Paper Deep Statistical Analysis and Revision Advice, a walkthrough of this 320-question report
More TMUA videos, including question worked solutions, on the YouTube channel and the 50 free practice questions page.

The exam, in numbers

  1. Across 2016–2023 the 320 official TMUA questions test 109 distinct skills out of a 123-skill syllabus dictionary, and a typical question chains about 3 skills at once (mean 3.2).
  2. At least 14 syllabus skills have never been tested in any official 2016–2023 paper, the full list unlocks free with the skill dictionary.
  3. The difficulty mix of all 320 questions is 33 questions at D2, 113 at D3, 124 at D4 and 50 at D5 (decision-load scale), mean difficulty D3.6.
  4. Papers have got harder: mean difficulty rose from D3.30–D3.45 in 2016–2019 to D3.85 in 2021, and D3.92 in 2023.
  5. The share of hard questions (D4 plus D5) climbed from 40% in 2018 to 75% in 2023; era on era, it is 48% in 2016–2020 versus 65% in 2021–2023.
  6. The five most-tested skills are counterexample construction (38 questions), necessary-and-sufficient conditions (35), differentiation (31), solving a quadratic (31) and index laws (30).
  7. The reasoning skills are also among the hardest: logical quantifiers average D4.1 and implication logic D4.0, while index laws average only D2.9.
  8. By topic, the most dangerous areas (highest share of D4–D5 questions, minimum 10 questions) are Logic of arguments (79%), Trigonometry (67%), Geometry (65%), Necessary & sufficient (65%) and Quantifiers (64%).
  9. Paper 1 is 99% single-answer (158 of 160 questions); Paper 2 is only 41% single-answer (65 of 160), the rest are reasoning formats: multi-statement (37), necessary-sufficient (22), proof-error identification (16) and counterexample (13).
  10. In recent papers (2021–2023) Paper 1's most frequent topics are Integration (about 3.7 questions per paper), then Algebra & functions and Trigonometry (about 3.3 each).
  11. The difficulty ramp is steep and real: in the recent era the first five slots average D2.9 while the last five average D4.7, and 81% of recent-era D5 questions sit in slots 14–20.
  12. Rare does not mean easy: of the 34 skills that have appeared at most twice, 17 arrived at D4 or D5 when they did, the pigeonhole principle struck at D5 in both 2019 and 2022, and 2023 saw a first-ever differential equation at D5 plus recurring-decimal conversion at D4.
  13. The Foundations (GCSE-assumed) strand is barely one question in ten but is examined at full TMUA difficulty: counting averages D4.4 across 16 questions and has appeared in all eight years, and polygon angles averages D4.5.
  14. Difficulty ratings are honest about uncertainty: 20 of the 320 sat on a boundary between two adjacent bands during review and are flagged in the Borderline section.
Difficulty is decision load, how much non-routine deciding a well-prepared student must do, rated D1–D5 (D1 never appears in a real paper). All ratings and classifications are Dr Grant's own assessments; this is independent analysis, not affiliated with or endorsed by the TMUA or its administrators.

The 40 most-tested skills

Frequency and mean difficulty across all 320 questions, most-tested first, 40 of the 109 tested skills. The full 123-skill dictionary is free to explore below: every skill with a plain-English definition, an EntryQ example question for most of them, the 14 never-tested skills, plus the TMUA Skills Compass PDF.
SkillHome topicStrandQuestionsMean difficulty
CounterexamplePrf1 Proof methodsReasoning38D3.61
Necessary and sufficient conditionsArg2 Necessary and sufficientReasoning35D3.83
DifferentiatingMM6 DifferentiationMain31D3.48
Solving a quadraticMM1 Algebra and functionsMain31D3.61
Index lawsMM1 Algebra and functionsMain30D2.87
Implication logicArg1 Logic of argumentsReasoning28D4.0
Logical quantifiersArg3 QuantifiersReasoning25D4.08
LogarithmsMM5 Exponentials and logarithmsMain24D3.5
Evaluating definite integralsMM7 IntegrationMain23D2.96
Exact trig valuesMM4 TrigonometryMain23D3.87
Expanding and simplifyingMM1 Algebra and functionsMain23D3.09
Completing the squareMM1 Algebra and functionsMain22D3.64
Signed areaMM7 IntegrationMain22D3.77
Solving trigonometric equationsMM4 TrigonometryMain22D3.68
Straight-line geometryMM3 Coordinate geometryMain22D3.59
Sketch and count intersectionsMM8 Graphs of functionsMain21D4.1
Linear simultaneous equationsMM1 Algebra and functionsMain20D3.45
Substitution to a quadraticMM1 Algebra and functionsMain20D3.35
Arithmetic progressionMM2 Sequences and seriesMain18D3.78
Divisibility and parityM2 NumberFoundations18D3.78
Stationary pointsMM6 DifferentiationMain18D3.89
Circle geometryMM3 Coordinate geometryMain17D3.76
DiscriminantMM1 Algebra and functionsMain17D3.65
OptimisationMM6 DifferentiationMain17D3.59
CountingM2 NumberFoundations16D4.38
Graph transformationsMM8 Graphs of functionsMain16D3.75
Identifying proof errorsErr1 Identifying proof errorsReasoning16D3.56
Trig graph propertiesMM4 TrigonometryMain16D3.5
ModulusMM1 Algebra and functionsMain15D4.33
Recurrence patternsMM2 Sequences and seriesMain15D3.47
SurdsMM1 Algebra and functionsMain14D3.07
Geometric progressionMM2 Sequences and seriesMain13D3.54
Solving quadratic inequalitiesMM1 Algebra and functionsMain13D3.62
Trig identitiesMM4 TrigonometryMain13D4.0
Careful inequality manipulationMM1 Algebra and functionsMain12D3.75
Binomial expansionMM2 Sequences and seriesMain11D3.18
Pythagoras and distanceM5 GeometryFoundations11D3.73
Common invalid deductionErr2 Common proof errorsReasoning10D3.3
Domain restrictionMM1 Algebra and functionsMain10D3.8
Linear-quadratic simultaneous equationsMM1 Algebra and functionsMain10D3.4

Difficulty over time

Difficulty here is decision load, how much non-routine deciding a well-prepared student must do, not step count and not topic hardness. The card below defines the five bands; the charts then track how the mix has shifted year by year.

How difficulty is graded (D1–D5)

One line per band. Colour chip matches the difficulty palette used in every chart below.
D1
Pure drill. One rehearsed technique, no decisions to make. Never appears in a real paper.
D2
One standard technique with at most one small twist. A named-theorem template counts as D2 even when it applies two conditions, extra plug-ins add volume, not depth.
D3
A couple of linked steps, or one non-obvious connection. The guided version of a hard idea: one genuine decision (which approach, one casework split, one validity check).
D4
Two named techniques genuinely chained, sustained multi-step work carrying a real idea, or one disguised insight a strong student misses at speed.
D5
The student has to originate the setup. Invent the representation or construction, combine more advanced skills across two or more topics, or sustain an extended reasoning chain. Time-to-see-it and scope-for-error weigh in too.
Adjacent bands are a judgement call. About 20 of the 320 sat genuinely on a boundary during review (see Borderline): a flagged rating sits on the line with one neighbouring band, so read it as that band or the single band it borders, not a whole band either way. The reliable signal is the overall pattern, the drift, the mix, the quotas, which is robust to any single borderline call.

Mean difficulty by year

Papers have drifted harder: both P1 and P2 trend upward after 2018, from a mean near D3.4 to nearly D4.

Difficulty mix by year (all 40 Qs/yr)

The D4+D5 share climbs from 40% of questions (2018) to 75% (2023), the paper now leans hard. White line = mean difficulty.

Strand & topic mix by primary subtopic tag

Each question is counted once under its primary subtopic. Main = Paper-1 maths (MM1–MM8), Reasoning = Paper-2 logic/proof (Arg/Prf/Err), Foundations = GCSE-assumed material that still headlines the odd question.

Strand share by year

Reasoning holds a steady ~¼ of the year (it lives almost entirely in Paper 2).

Strand split by paper

P2 is roughly half Reasoning, half maths re-used as proof material.

Topic × year heatmap

Cell = number of questions whose primary subtopic sits in the topic that year (both papers). Hover for the P1/P2 split and mean difficulty.

Question formats formats

Beyond plain single-answer MCQ, TMUA leans on reasoning formats, multi-statement (“which of I, II, III…”), necessary/sufficient, spot-the-error proofs, counterexamples. These are heavily concentrated in Paper 2 and are, on average, harder than single-answer questions.

Format mix by year Per question

Format profiles Per question

Syllabus map Strand › Topic › Subtopic icicle

Width of each block = how many of the 320 questions centre on that subtopic (Coverage, counted for every primary tag). Colour = mean difficulty of those questions (cool → hot). This is the single-picture answer to “what actually comes up, and how hard is it when it does”.
≈D3 when it appears≈D4≈D4.7+

Skill dictionary plain-English glossary

A plain-English definition for every skill on the syllabus, the 109 the papers actually test (most with an EntryQ example question showing it in action) and the 14 that have never come up. Read it as a glossary before the skill charts below, the charts tell you how often each skill comes up and how hard it tends to be, and this tells you what each one actually means.

Unlock the full 123-skill dictionary, free

Free unlock: every skill, plus the printable revision guide

Every skill on the TMUA syllabus with a plain-English definition and, for most skills, an EntryQ example question showing it in action, the 109 skills the papers actually test, plus the 14 syllabus skills that have never come up, so you know exactly what to prioritise, and what can safely wait until last.

Something went wrong, please try again (or email [email protected]).
Instant unlock, right on this page: every skill, interactive, most with an EntryQ example question and its answer. A printable review copy (the TMUA Skills Compass PDF) is included. You’ll also get new TMUA practice questions by email and news of the EntryQ app (about 2,500 questions, open to pass holders). No spam, unsubscribe anytime. One unlock per student, for personal use only, not for schools or tutoring companies. Full terms.

Skills distinct skills

The skill layer is where “what to revise” lives. The scatter answers the two student questions at once: how often does a skill appear (x) and how hard are the questions that use it (y). Bubble = one skill, sized by frequency, coloured by home strand.

Frequency × difficulty, every skill All tags

Top-right = common and hard (priority revision). Bottom-right = common but routine (bankable marks). Hover any bubble.

Top-40 skill × year heatmap All tags

Appearances per year (a question can use several skills). Hover for exact counts.

All skills, sortable All tags

Click a column header to sort; type to filter. “Hard share” = % of its questions rated D4 or D5.

🔒 Free with sign-up

The full skills analysis: the frequency × difficulty scatter of all 109 tested skills, the top-40 skill × year heatmap, and the complete sortable table.

Something went wrong, please try again (or email [email protected]).
One sign-up unlocks every section, plus the printable revision guide and news of EntryQ. One unlock per student, terms.

Skill co-occurrence which skills get tested together All tags

TMUA rarely tests one skill in isolation, so we count how often two skills are needed in the same question (all tags, primary and secondary). Start with the plain ranked list of the commonest pairings; open the network explorer only if you want to see the clusters.

Top 15 skill pairings All tags

The pairings real papers repeat most. Bar colour marks the strand pair; number = times the two skills shared a question.

🔒 Free with sign-up

Which skills get tested together, the top pairings and the interactive co-occurrence network. Hard questions chain skills; this is the map of the chains.

Something went wrong, please try again (or email [email protected]).
One sign-up unlocks every section, plus the printable revision guide and news of EntryQ. One unlock per student, terms.

Paper anatomy difficulty by question position

Where in the paper do the hard questions sit? We split the eight years into two eras: the base era (2016–2020, 200 questions) and the recent era (2021–2023, 120 questions). The clean difficulty break in the data falls after 2019 (2020 is transitional), so we fix the recent era at 2021–23 for consistency with those sections. The recent era is drawn boldest throughout, because that is the exam students actually face.

Paper 1, difficulty by slot

Mean difficulty at each slot Q1–Q20. Recent era (bold) runs above base almost everywhere, the whole ramp lifted.

Paper 2, difficulty by slot

Same view for Paper 2. Its back third climbs to a solid D4–D5 in the recent era.

Difficulty heat strip, base vs recent

Cool → hot mean difficulty per slot. Two rows per paper: base era on top, recent era below, so the ramp change is visible at a glance.

Where D5s have landed

Each mark is a slot that has ever produced a D5. Filled = a D5 landed there in the recent era (2021–23); hollow = D5 only in the base era. The traps have crept forward and clustered in the back third.

Skills over time what to prioritise, year by year

This is the revision compass. It disaggregates all 100+ skills across the eight years without ever crowding them onto one screen: start from the big picture (strands, then topics), then drill into a filtered skill heatmap. Switch between counting only each question's primary tag or all its tags (a hard question spans several subtopics at once).

Level 1, headline shifts auto-computed

The five things the data says loudest right now. “Recent” = 2021–23, “early” = 2016–18.

Strand share of the paper, by year Primary tag

Every question counted once (primary) or once per strand it touches (all). Hover a band for the exact count.

Level 2, topic momentum 25 topics Primary tag

Each row is a topic; the strip shows how many questions it took each year (darker = more). Click a topic to focus the skill explorer below on it.

Level 3, skill explorer Primary tag

The full skill × year grid, kept navigable by only ever drawing the rows you ask for. Sorted by total, most-tested at the top.

🔒 Free with sign-up

The revision compass: strand, topic and skill trends year by year, filterable by paper and topic, so you weight your revision to the exam as it is now.

Something went wrong, please try again (or email [email protected]).
One sign-up unlocks every section, plus the printable revision guide and news of EntryQ. One unlock per student, terms.

Where the hard questions live difficulty × topic

Not every topic is equally dangerous. This ranks the 25 topics by their share of D4–D5 questions, so you can see which topics reliably produce the paper's toughest marks, and which are usually gentle. Bar = the topic's D2→D5 mix; number = % of its questions rated D4 or D5.

🔒 Free with sign-up

The danger-zone ranking: which topics reliably produce the D4–D5 questions, and which are usually gentle.

Something went wrong, please try again (or email [email protected]).
One sign-up unlocks every section, plus the printable revision guide and news of EntryQ. One unlock per student, terms.

Borderline ratings 20 flagged

Difficulty here is a judgement of decision-load, and 20 of the 320 sat genuinely on a boundary between two ratings during review. They are listed for honesty: each is shown with the two adjacent bands it genuinely sits between (the “sits between” column). Everything else was a clear call.

Unlock terms please read

The report unlock is free to everyone who signs up, with no expiry. It is for personal use and is not a resource to be pooled, shared or redistributed. The EntryQ app is a separate paid product; its passes and prices are on the EntryQ page.
  • The unlock may not be shared, redistributed, resold, sub-licensed, or used by a school, tutoring company, agency or other organisation to provide access on behalf of multiple students or staff. Each student who wants the unlocked content should sign up individually with their own email address.
  • Schools, tutoring companies and other organisations wanting to provide this to a group of students should contact [email protected] to arrange a separate group or institutional licence.
  • All data, difficulty ratings, skill tagging, findings, charts and written analysis on this page are original work by Dr Nicky Grant and are protected by copyright. Reproducing, republishing or otherwise reusing them, in whole or in part, in any format, requires prior written permission (see the notice at the top of the page).

Privacy & pupil data EntryQ app

The EntryQ app is built to minimise personal data. Here is the short version, with the full notice one click away.
  • You need an account to use the app (an email address and a password). Your email and sign-in records are held by Supabase for EntryQ, in Ireland; the notice lists the other providers.
  • Your practice is stored on your device. Account backup is on by default and keeps a copy of your practice in your account so you can carry on from another device; you can switch it off in Settings. Whatever that switch says, to protect the question bank, the service records each question it sends you and, when you ask for an answer, the option you chose.
  • Your Plan and Progress are worked out on your device and are not sent to EntryQ. Letting EntryQ use your practice data for pooled figures and research is optional and off until you turn it on.
  • No data selling, no adverts, and no advertising or product-analytics trackers. Under the proposed school arrangement, schools would not supply pupil lists or see individual results.
  • The app is for people aged 16 or over. A short summary is in the EntryQ privacy summary; the full notice, including your rights and how long data is kept, is shown inside the app before you create an account, and we will email a copy on request.

Turn this analysis into a plan

I tutor TMUA candidates one-to-one, using exactly this question-by-question data to target the skills that decide your score. The first step is a free 30-minute consultation.

Book a free consultationSit the free timed exam

Free 30-min consultationNo obligation · reply within one business day
Book now