How to use past papers properly

Past papers are the highest-fidelity study material that exists for an exam. They're written by the people who write the exam, in the format of the exam, at the difficulty of the exam. And they're routinely used in the way that extracts the least from them: opened alongside the mark scheme, worked through in fragments over an evening, "checked" rather than marked.

There's a method that gets much more out of them. It takes longer per paper, and you'll do fewer papers as a result. That trade is worth making.

Before: use them as a map

Before you start revising a subject in earnest, spend an hour reading — not doing — three or four papers. You're looking for structure, not answers.

• Which topics show up every single year? Those are the guaranteed marks and they should get proportionate time.
• How are the marks distributed across the paper? A section worth 40% deserves more than a quarter of your preparation.
• What do the command words demand? "State", "explain", "evaluate" and "justify" are asking for genuinely different things, and mark schemes are strict about it.
• What does a full-mark answer look like — how long, how structured, how much working shown?

An hour spent here changes what you revise. It's the cheapest prioritisation available.

During: make it a real exam

The conditions matter more than people expect, because you're practising retrieval under load, not just answering questions.

Closed book. Non-negotiable. A paper done with notes open measures your ability to look things up, which won't be tested.

Timed, with the real allowance. Time pressure is a large part of what makes exams hard, and it's the component you can only practise by imposing it. If you can't manage the whole paper, time a single section properly rather than doing the whole thing untimed.

In one sitting. Splitting a paper across three evenings removes the fatigue, and fatigue in the last forty minutes is a real and trainable problem.

Guess rather than skip. If you'd write something in the real exam, write it. You want the marked script to reflect what you'd actually produce, including the half-remembered attempts.

After: the part that matters

Budget as long for marking as for sitting. If the paper took ninety minutes, the review is not a ten-minute job.

1. Mark it like a stranger wrote it

Use the official mark scheme and be unkind. "I basically said that" is not a mark. If the scheme wanted the word because followed by a mechanism and you wrote a description, that's zero, and knowing it's zero now is worth a great deal more than the comfort of awarding yourself the point.

Read the scheme properly while you're there. Mark schemes are the clearest available statement of what the examiner values — where the marks sit in a long answer, which keywords are required, how much working must be shown. That information is worth having independently of your score.

2. Classify every lost mark

This is the single highest-value habit in this guide. For each mark you didn't get, write one of:

The reason this matters: these five failures look identical on the score line and need completely different responses. Someone losing fifteen marks to timing and someone losing fifteen marks to gaps should spend the next week doing entirely different things. Without the classification, both will just "revise more".

3. Redo the ones you got wrong — properly

Reading the model answer is not enough; it produces exactly the fluency illusion this whole exercise exists to defeat. Close the scheme and rewrite the answer in full from memory. If you can't, you haven't fixed it yet.

Then, a week later, do those same questions again cold. Anything you fail twice is a genuine weak point, and now you know exactly which they are.

4. Keep the record

One paper is noise. Four or five papers, with topics and causes logged, is a diagnosis: "I lose most of my marks on the extended 6-markers, mostly to running out of time, mostly in the last section." That sentence is worth more than any individual score, and you can only get it by writing things down as you go.

Sequencing and supply

There are only so many past papers, and they aren't equally valuable. A rough order:

• Early — oldest papers, or ones from a previous specification. Untimed, topic by topic, as practice questions rather than exams.
• Middle — full papers, timed, one every week or two, with the full review each time.
• Last fortnight — the two most recent papers, saved deliberately, sat under proper conditions. These are your best simulation and you only get to use them once.

If you run out, examiners' reports are underrated: they say in plain language what most candidates got wrong and why, which is a fair prediction of what you'll get wrong too.

If you take one thing

Next paper you mark, write a cause next to every lost mark before you look at your total. Five minutes of extra work, and it converts a score into a plan.