Past papers: the right way and the three wrong ways to use them

The short answer: a past paper is a diagnostic tool, not a revision activity. Its value is almost entirely in what you do in the hour after you finish it — marking it honestly, working out why each lost mark was lost, and doing something about the pattern. Students who treat the paper itself as the revision get a score, a vague feeling of having worked hard, and no improvement. I've watched students grind through fifteen papers and move three marks. I've watched others do six properly and move two grades. The difference is never the papers.

Wrong way one: doing papers before you can do the topics

This is the most common, and it feels the most virtuous. A student who hasn't properly learned integration by parts sits a full paper, scores 40%, and concludes they need to do more papers. What they've actually done is spend two hours confirming something a five-minute conversation would have told them.

Past papers are a finite resource. Once you've seen a question you can never again use it to find out whether you could have done it cold, so burning papers as a substitute for learning content is expensive in a way that isn't obvious until March, when you've run out.

The rule I give: don't sit a full paper on a topic you couldn't teach to someone else. Learn it from your notes, your teacher, or a worked example, do a set of targeted questions on that topic alone, and only then let it appear in a full paper. Full papers are for testing whether your knowledge survives contact with a mixed, timed, unpredictable environment — not for acquiring it.

Wrong way two: marking your own work generously

This one costs more marks than anything else on the list, and almost nobody believes they do it.

Here is what it looks like in practice. Your answer is 3.72, the mark scheme says 3.27, you spot that you transposed two digits, think "silly mistake, I knew how to do it", and award yourself the marks. You leave a question half-finished, read the solution, understand it instantly, and mark it as basically right. You write a proof that reaches the right conclusion through a step that doesn't follow, and tick it because the last line matches.

An examiner will do none of those things. The mark scheme isn't a guide to whether you understood — it's a list of specific things that must appear on the page. Mark yourself the way an examiner would and your scores will drop noticeably overnight. That drop is the most useful information you'll get all year: it's the gap between what you think you can do and what you'd actually be awarded.

One rule makes this easier: when you're genuinely unsure whether something earns the mark, assume it doesn't. Being harsh on yourself in October costs you nothing.

Wrong way three: finishing the paper and moving on

The paper is finished, it's marked, the score goes in a spreadsheet, and the next paper starts. This is the version that feels most like a proper revision programme and produces the least improvement, because nothing has been fixed.

What's missing is the error log. Every lost mark goes into a list with two columns: what went wrong, and what category it belongs to. Over four or five papers, the categories stop being random and start being a short, specific, deeply annoying list — and that list is your actual revision plan. Most students discover they have three or four recurring failure modes, not thirty. Sign errors when expanding brackets. Not reading "give your answer to three significant figures". Setting up mechanics problems without drawing the forces. Losing the constant of integration.

You cannot revise "being bad at maths". You can absolutely revise "I lose four marks a paper to sign errors", and it takes about a fortnight.

The right way: the protocol

This is what I ask students to do, and it takes roughly twice as long as the paper itself. Budget for that. One paper properly digested per week beats three skimmed.

• Sit it in one sitting, phone in another room. A paper done across three evenings tells you nothing about whether you fade in the last half hour — and fading in the last half hour is a real, fixable problem.
• Mark it yourself, harshly, in a different colour. Not a parent, not a tutor, and not the mark scheme read casually afterwards. You have to see your own work through an examiner's eyes; that skill is worth marks on its own.
• Categorise every lost mark. Content you didn't know, method you knew but executed badly, careless slip, or misread question. Those four need four different responses, and treating them all as "revise harder" is why nothing changes.
• Redo the questions you got wrong, from blank, two days later. Not the same evening while the solution is fresh in your head — that tests your memory of the answer, not your ability to produce it. Two days later, from a blank page, is the test that counts.
• Look at the error log before the next paper. Five minutes. It's the only part of this that's optional-feeling and non-negotiable in practice.

When to start timing yourself

Later than most students think, and earlier than most students actually do.

Working untimed at first is sensible: you're finding out whether you can do the mathematics at all, and a clock just adds noise. But at some point the constraint is the exam, and an untimed 85% that would have been a timed 60% is a comforting fiction. My sequencing: untimed until you're reliably getting the marks, then timed with generous overrun allowed, then strictly timed from the mock period onwards. By the spring of Year 13 every paper should be under exam timing, because what you're training by then is not knowledge but speed and composure.

Which papers, and in which order

Use your own board's papers, and check which board you're on before you start rather than assuming. The differences between boards are real — question phrasing, how the applied content is split across papers, and the house style of the longer questions all vary — and practising against the wrong board's papers builds slightly wrong instincts. Our hub has the papers organised by board: Edexcel, AQA and OCR, with Further Maths and International A-Level on the main past papers page.

For order: work backwards from the most recent, but hold two or three of the newest papers back, untouched, for the final month. You want a couple of genuinely unseen papers left when you're actually match-fit, so you get a clean read on where you stand.

How many papers is enough

Fewer than you'd guess, done properly. For a Year 13 aiming at an A or A*, ten to twelve full papers across the spring — one a week from January, picking up after Easter, each with the full mark-and-redo cycle — is a serious programme. A student claiming thirty is either not marking them properly or not fixing anything; there aren't enough hours in the week to do thirty well alongside three other A-Levels.

If you're rebuilding after a bad mock the ratio shifts: more targeted topic practice, fewer full papers, until the content is actually there. The four-week mock plan covers that sequencing in more detail, and if you're going into Year 12 rather than Year 13, the summer preparation post is the more useful starting point.

If you're a parent reading this

The single most useful thing you can do is stop asking about scores. "What did you get?" turns every paper into a verdict, and a student who fears the verdict will quietly start marking generously to make the conversation easier. Ask what went into the error log instead. It's harder to fake, and it makes the paper a tool rather than a test — which is the entire point.