The direct answer
How to revise maths for exams: list the problem types the exam can ask, learn one type at a time by explaining a worked example out loud, then practise mixed sets under time and log every error by the step that went wrong. Maths is revised by doing problems, not by reading about them.
How to revise maths for exams starts with doing, not reading
You read the chapter, the worked examples make sense, and you feel ready. Then the exam gives you a problem you have not seen before, and the method you “knew” will not come. You stare at the question, try the first step, and it is wrong — so the rest is wrong too.
That is the maths trap. Reading a worked example feels like understanding, but understanding a solution is not the same as producing one. Maths exams test whether you can choose the right method for a new problem and carry it through without mistakes. The only practice that trains that is solving problems yourself, from a blank page.
The research is blunt about the mixing part. In Rohrer and Taylor’s 2007 study, students who practised four kinds of volume problems mixed up scored 63% on a test a week later, against 20% for students who practised them blocked by type — even though the blocked group did far better during practice itself, 89% against 60%. Fourth-graders showed the same pattern: 77% against 38% a day later.
The system in short
- List the problem types the exam can ask. One line each, from past papers.
- Learn one type: study a worked example, then explain each step out loud in your own words.
- Do five problems of that type from a blank page, no notes, and mark them with feedback.
- Mix the types up: ten to fifteen problems, random order, timed.
- Log every error by the step that went wrong, and retest the logged problems after a gap.
Step 1: List the problem types
Open the past papers and the syllabus. Write down every problem type the exam can ask, one line each. “Solve a quadratic by factorising.” “Find the volume of a cone.” “Differentiate a product.” Be exact — a type is a kind of problem with a method, not a chapter heading. Most maths exams test six to twelve types: fewer means you missed some, more means sub-steps.
This list is the whole revision plan. Cross types off only when you can solve them mixed up, not just blocked.
Step 2: Learn one type by explaining a worked example
Pick one type from the list. Find a worked example in your notes or textbook. Read it once, slowly, then close the book and explain each step out loud, in your own words: what the step does, and why it comes next.
“First I move everything to one side, because I need the equation equal to zero before I can factorise.” That sentence is the learning. If you cannot say why a step comes next, you have memorised a sequence, not understood a method — and a sequence breaks the moment the exam changes one number.
This is self-explanation. In Chi’s 1989 study, eight physics students studied worked examples while thinking aloud: the four who later solved problems best had generated 15.3 explanations per example, against 2.8 for the weakest, scoring 82% against 46%. The study was small, so it shows a pattern, not a proof — but a 2018 meta-analysis found prompting self-explanation gave a moderate average benefit across 64 reports.
One type per session at most. Learn it properly, then move on. Revision is the last few weeks — the full weekly system for learning maths through the term is the study system for maths students.
Step 3: Five problems of that type, blank page, no notes
Now do five problems of the type you just learned, from a blank page, with no notes open. Set a timer — the exam’s time per question is a good guide — and do not look at the worked example while you work.
Mark them as soon as you finish, with feedback. Feedback matters: Rowland’s 2014 meta-analysis found testing yourself beat restudying, and the effect was larger with feedback (g = 0.73) than without it (g = 0.39). For every wrong answer, find the exact step where it went wrong. Not “I got it wrong” — “I factorised wrong at step 3, because I missed the negative sign.” Fewer than three out of five right means the type is not learned yet.
Step 4: Mix the types up in timed sets
Once you can do each type alone, stop practising them one at a time. Build mixed sets: ten to fifteen problems, drawn at random from all the types you have learned, under timed conditions.
This is the step that makes the exam feel familiar. In the exam, nobody tells you which type a question is — you have to work that out from the question itself, then choose the method. Taylor and Rohrer’s fourth-graders who mixed their practice scored 77% a day later against 38% for the blocked group, because mixing forced them to choose the right procedure for each problem instead of repeating the last one. Mixed sets will feel worse than blocked sets, and your score will drop at first.
Step 5: Log the errors by step, and retest after a gap
After every set, write down each error with the step that went wrong. Keep it short: “Q4, quadratic, step 3, missed negative sign.” Sort the log by step, not by question — if three errors are all “step 3, sign error”, you have one problem, not three.
Retest every logged problem after a gap of two to three days, mixed in with new problems. Do not retest the same day: you will remember the answer, not the method. A two-to-three-day gap is a practical starting schedule, not a research rule.
When a logged problem comes back right in a mixed set, cross it off. When it comes back wrong at the same step, it goes back to step 2: explain a worked example of that type again, out loud, before you try it a third time.
A worked example: a Monday evening, three weeks before the exam
You have a maths exam in three weeks. Twelve problem types on the list. Monday evening, two hours.
6:00–6:10 — Pick the type. Quadratics by factorising. One worked example, read once.
6:10–6:25 — Explain it out loud. Each step, why it comes next. “Move everything to one side, because I need zero on one side before I factorise.” Three explanations written down.
6:25–6:55 — Five problems, blank page. No notes. Timer on. Four right, one wrong — step 3, missed a negative sign. Logged.
6:55–7:10 — Break. Walk, water, no phone.
7:10–7:50 — Mixed set. Twelve problems, drawn at random from the six types learned so far. Nine right. The three wrong ones: two sign errors, one wrong method choice.
7:50–8:00 — Log and plan. Total today: 17 problems, 4 wrong, 3 of them sign errors at step 3. Tomorrow’s ten minutes: three sign-error problems, explained out loud, before anything new.
That is two hours, and it produced something a day of re-reading never does: a list of exactly which step breaks, and a plan for tomorrow that is ten minutes long.
When it goes wrong
You read worked examples and call it revision. Reading a solution is not solving a problem. The rule is one read, then close the book. If you cannot do the problem from a blank page, you have not learned the type yet.
Your mixed sets score lower than your blocked sets, so you go back to blocked. That is the trap the research warns about. Blocked practice scored 89% during practice and 20% a week later in Rohrer and Taylor’s study. Judge the system by the mixed score after a week, not by the blocked score today.
The same step keeps going wrong. Three sign errors, three sessions in a row. That is not bad luck — it is a gap in step 2. Go back and explain a worked example of that type out loud, slowly, saying why each step comes next. Then do five problems of that type alone before you mix again.
You do twenty problems and mark none of them. Then you have practised, but you have not learned. Feedback is what turns practice into learning — Rowland found the testing effect nearly doubled with feedback. Mark every problem, and write the step that went wrong for every error.
Weekly review checklist
Copy this list and use it each week:
- How many problem types can I now solve in a mixed set, and which ones still break?
- What is my most common error step, and how many times did it appear this week?
- Did I retest every logged problem after a gap, mixed in with new ones?
- What one change will I make next week: relearn a type, more mixed sets, or slower marking?
Downloadable infographic spec: one vertical image, title “The maths revision system in five steps”, with the five steps — list the problem types, explain a worked example, five problems blank page, mix the types up, log errors by step — and the rule, one read, then close the book, in Thynq cream, Cloud Purple, Dark Purple and pink, with the cloud mascot. Text taken word-for-word from this article.
What this system is not
It is not a promise of a grade, and it is not a claim that doing problems is the only way to learn maths. Understanding the method matters — that is what step 2 is for. But understanding without doing is not exam preparation. The exam asks you to solve new problems under time, and the only way to get good at that is to solve problems under time, mixed up, with feedback.
Sources
- Rohrer, D. & Taylor, K. (2007). The shuffling of mathematics problems improves learning. Instructional Science, 35(6), 481–498. Interleaved practice scored 63% vs 20% blocked a week later; blocked did better during practice (89% vs 60%).
- Taylor, K. & Rohrer, D. (2010). The effects of interleaved practice. Applied Cognitive Psychology, 24, 837–848. Fourth-graders: interleaving scored 77% vs 38% a day later.
- Chi, M. T. H., et al. (1989). Self-explanations: how students study and use examples in learning to solve problems. Cognitive Science, 13(2), 145–182. Eight physics students; best solvers gave 15.3 explanations per example vs 2.8, scoring 82% vs 46%.
- Bisra, K., et al. (2018). Inducing self-explanation: a meta-analysis. Educational Psychology Review, 30(3), 703–725. Prompting self-explanation gave a moderate average benefit (g = 0.55).
- Rowland, C. A. (2014). The effect of testing versus restudy on retention: a meta-analytic review of the testing effect. Psychological Bulletin, 140(6), 1432–1463. Testing beat restudy (g = 0.50); feedback raised the effect (g = 0.73 with, 0.39 without).
- Dunlosky, J., et al. (2013). Improving students’ learning with effective learning techniques. Psychological Science in the Public Interest, 14(1), 4–58. Only practice testing and distributed practice rated high utility; self-explanation and interleaved practice moderate.
- Bjork, R. A. (1994). Memory and metamemory considerations in the training of human beings. In Metcalfe & Shimamura (eds.), Metacognition: Knowing about knowing, MIT Press. Performance now is not learning later.
For the full exam system this sits inside, start with how to revise for exams without cramming. For timed written practice, the past paper system for exams is the companion piece, and if you want the recall-first habit in general, the retrieval practice system for students is the same idea applied to any subject.
Questions
Should I revise maths by reading my notes or doing problems?
Doing problems. Reading notes feels like learning, but maths is a doing subject — the exam asks you to solve problems you have not seen before. Read a worked example once, then close the book and do a similar problem yourself.
What do I do when I get stuck halfway through a problem?
Write down what you know and what the question is asking for, then look at the step you are stuck on. If you cannot name the step, that is the gap — go back to the worked example for that one step only, then try the problem again from the start.
How many problems should I do per session?
Ten to fifteen in a mixed set is enough for one session. More than that and you stop thinking and start copying the pattern of the last problem. Every wrong answer needs a one-line note: which step went wrong, and why.
Why do I do well in practice but badly in the exam?
Usually because practice was blocked — one problem type at a time — and the exam mixes them. Blocked practice feels fluent because you repeat the last method. Mixed practice feels worse during the session and works better on the test a week later.
