A-Level Maths past papers: where to find them and how to actually use them
Past papers are the single most effective A-Level Maths revision resource, but only when used as diagnosis rather than dress rehearsal. Do a paper (or half paper) under timed conditions, mark it against the official mark scheme, log every dropped mark with its cause, then fix those specific topics before the next paper. All boards publish papers, mark schemes and examiner reports free on their websites — Edexcel at Pearson, AQA and OCR on their own qualification pages.
Every student revising A-Level Maths ends up doing past papers. The difference between the ones who improve and the ones who just accumulate completed PDFs is entirely in what happens after the paper is finished. This is the system that works, and where the papers actually live.
Where the official papers are
| Board | Spec code | Where | What's free |
|---|---|---|---|
| Edexcel | 9MA0 | Pearson qualifications site → Mathematics → past papers | Every paper, mark scheme and examiner report from 2018 on, plus specimen papers |
| AQA | 7357 | AQA site → A-Level Mathematics 7357 → assessment resources | Papers, mark schemes, specimen sets and examiner reports |
| OCR A / MEI | H240 / H640 | OCR site → the relevant specification → assessment | Papers, mark schemes, examiner reports and practice sets |
The method that makes papers work
- Do the paper (or a 60-minute section) under real time pressure. No notes, no pausing, phone in another room.
- Mark it yourself with the official mark scheme, and be an examiner, not a friend. Method marks require the method to be visible.
- For every dropped mark, write one line on the cause. 'Forgot the −c', 'didn't know R-form', 'misread "hence"' — each implies a different fix.
- Fix only the things on that list. Not the whole topic — the specific step.
- A week later, redo just those questions from a different paper. Same questions re-done prove nothing.
The log is the part everyone skips and the part that matters most. After three or four papers, patterns emerge that no single paper shows: a student who loses two marks every series to dropped negatives has a different problem from one who loses eight to statistics, and only the log tells you which you are.
Which order to do them in
- Phase one — topic work: for the topics you already know are weak, use topic-sorted questions (Physics and Maths Tutor organises every past question by topic) before touching a full paper.
- Phase two — timed sections: 60-minute halves of 2019–2023 papers, one or two a week, building stamina and speed.
- Phase three — full mocks: the two most recent series, sat back-to-back days as real mocks, in the final month.
- Specimen papers (SAMs): useful for extra volume but slightly unrepresentative — they were written before any real exam existed. Do them mid-phase, not last.
How many marks a paper is really worth
A 100-mark paper marked honestly and followed up properly is worth more than five papers done and filed. The follow-up — the log, the fixes, the re-test a week later — is where the learning happens. If you only have time for one of them, do fewer papers properly.
A student drops 6 marks on a Paper 3 statistics question: they wrote P(X = 14) instead of P(X ≥ 14) in a binomial test, then concluded correctly for the wrong probability.
- Log line: 'used point probability instead of tail — knew better, rushed'.
- Fix: the rule is 'the test statistic goes in the direction of the alternative hypothesis' — H₁: p > means an upper tail.
- Re-test: find two hypothesis-test questions from a different year's Paper 3 and do only the probability-setup step.
- A week later, one more question cold. If the tail is right, the issue is closed; if not, it goes back on the list.
One logged cause, two targeted questions, one confirmation — that's the entire loop. Six marks recovered for next time.
Common past-paper mistakes
- Marking generously — awarding method marks for method you meant but didn't write. The examiner can only mark what's on the page.
- Doing papers untimed until April, then discovering the real problem was speed.
- Hoarding papers 'for when I'm ready' — the gaps they reveal get found too late to fix.
- Only re-doing the questions you got wrong, not the question type. One context, learned, doesn't transfer.
- Ignoring examiner reports while doing their papers — the complaints are the same every year.
The checklist
- I know which board and spec code I'm sitting, and I've downloaded that board's papers only.
- I have a dropped-marks log and every entry has a cause, not just a score.
- Every full paper I've done was timed, no-notes, marked against the official scheme.
- I've re-tested every fixed weakness on a different paper a week later.
- The two most recent series are saved, untouched, for final mocks.
- I've read the last two examiner reports for each of my three papers.
The one step in the loop that papers can't do for you is the explanation — when the mark scheme's working doesn't make sense, you need it explained differently. That's the specific moment Lumi exists for: upload the question, point at the line, and it shows the step and then hands you a similar one to confirm it stuck.
What to take from this
- Start papers months before the exam — they are for finding gaps, not confirming you are ready.
- Official sources only: Pearson (Edexcel), AQA and OCR all host free papers, mark schemes and examiner reports.
- The mark scheme is a revision resource, not just a score sheet — it shows the method marks.
- Log dropped marks by cause: 'didn't know it', 'misread', 'ran out of time' each need a different fix.
- Save the two most recent series as your final timed mocks; mine 2018–2023 for practice.
Questions people also ask
All of them from the current spec (2018 onwards) is the ideal — that is eight-plus full sets per paper code. If time is short, five or six done properly with full follow-up beats twelve done and filed.
The content is nearly identical, so yes for extra volume — but do your own board first to learn its command words and style. Edexcel's large data set and AQA's multiple-choice sections have no equivalent on the others.
Official past papers are free and real. Predicted papers are reasonable extra volume once those are exhausted, but they are guesses at question style — never substitute them for the real thing.
Roughly, but grade boundaries move each series and mock conditions inflate scores — quiet room, no pressure, familiar building. Treat your paper score minus five to ten percent as the honest estimate.
Timed sections from January of Year 13; full papers by Easter. If you are reading this earlier, start earlier — papers are for finding gaps, and gaps found in autumn are fixable.