
How to Revise for a Maths Exam: The Problem-First Method
Maths exams punish one study habit more reliably than any other: reading notes instead of solving problems. The research on procedural skill development, including work by Rohrer and Taylor (2007) published in Psychological Science, consistently shows that doing problems under varied conditions produces far better performance on delayed tests than restudying worked solutions. Reading through your calculus notes the week before an exam builds familiarity, not fluency. The exam asks you to execute, and execution degrades without practice in a way that conceptual knowledge does not.
Why Maths Revision Is Different From Every Other Subject
Revising for maths requires different cognitive work than revising for any content-heavy subject. The distinction separates students who plateau at "I understand it when I see it" from those who can reliably produce correct answers under time pressure.
The Re-Reading Trap: Why It Feels Productive But Fails
Reading through a worked solution to a calculus problem produces recognition, not recall. The solution looks logical, each step follows from the last, and your brain registers the process as known. That feeling of familiarity is the trap. Dunlosky et al. (2013), reviewing ten learning techniques across 119 studies, rated re-reading as having “low utility” for learning, partly because the feeling of fluency it produces is not matched by actual performance on retrieval tasks.
In maths, the failure mode is sharper than in other subjects. A history student who re-reads might recall enough to construct an argument. A maths student who re-reads can follow the logic of a worked example perfectly but cannot reproduce the first step without the example in front of them. The skill being tested, generating the solution under time pressure, was never practiced.
Maths Tests Performance, Not Just Knowledge
Understanding the method for integrating by parts and reliably integrating by parts in an exam are two different competencies. The first is declarative knowledge: you can describe what to do. The second is procedural skill: you can do it accurately and quickly under conditions where errors compound and time pressure is real.
Procedural skill degrades without practice. A student who practised integration by parts thoroughly during the taught course but did not revise it for eight weeks will be slower and less accurate than one who practised it ten days ago. Reading the method again does not restore the speed. Only performing the method does. That is why maths revision is primarily about volume of problem-solving, not volume of review.
How to Build a Problem Bank for Maths Revision
A problem bank is a curated collection of exam-style questions organised by topic and difficulty. Building one before you start revision ensures that when you sit down to practise integration by parts or hypothesis testing, you spend your time on problems rather than searching for them. The structure of the bank also forces an honest audit of which topics your course actually covers.
Where to Source Good Problems
Past exam papers are the highest-yield source. They match your exam's question style, difficulty, and marking conventions exactly. Your department or faculty will hold past papers, often accessible through your university library portal or department page.
Textbook exercise sets provide additional volume, particularly for topics where recent past papers have few examples. For university-level mathematics, OpenStax free mathematics textbooks (Calculus, Statistics, Precalculus) provide well-organised problems at levels matching typical first and second-year courses. MIT OpenCourseWare mathematics courses include problem sets and exams with solutions for single-variable calculus, linear algebra, probability, and differential equations, all freely accessible. Both are legitimate academic sources, not content mills.
Organising Problems by Topic and Difficulty
A flat list of problems is not a problem bank. Organise your bank so you can pull problems by topic and difficulty in under 30 seconds. A simple spreadsheet with columns for topic, difficulty (1-3), source, and completion status works better than elaborate note-taking systems. You want the overhead of using the bank to be zero so revision sessions start immediately.
| Topic area | Problems collected | Difficulty spread | Status |
|---|---|---|---|
| Integration by parts | 18 problems | 6 easy / 8 medium / 4 hard | In progress |
| Hypothesis testing | 22 problems | 8 easy / 9 medium / 5 hard | Not started |
| Matrix operations | 15 problems | 5 easy / 7 medium / 3 hard | Complete |
| Differential equations (first order) | 20 problems | 7 easy / 9 medium / 4 hard | In progress |
Example problem bank structure. A rough difficulty spread of 40% easy, 45% medium, 15% hard matches the profile of most university maths papers.
Target a rough spread of 40 percent straightforward problems, 45 percent medium problems requiring two or three steps, and 15 percent harder problems. Do not fill the bank with hard problems only. Easy and medium problems build fluency and speed, which matter as much as the ability to handle hard questions.
What Is an Error Log and How Do You Run One?
An error log records every problem you get wrong or cannot complete, capturing the exact nature of the error so you can fix the right thing. The log turns failed attempts into the most useful data in your revision, because repeated errors in the same place reveal genuine gaps rather than careless slips.
The Three-Column Error Log Workflow
Three columns cover everything you need. The first column names the problem type precisely (not just "integration" but "integration by parts where u-substitution would fail"). The second column describes the exact error: the specific step where the working went wrong, the conceptual misunderstanding behind it, or the procedural slip. The third column states the correct approach in your own words.
| Problem type | Exact error made | Correct approach (own words) |
|---|---|---|
| Integration by parts: ln(x) type | Set dv = ln(x)dx instead of u = ln(x). Could not integrate ln(x) directly. | Always set u = ln(x) because its derivative 1/x simplifies. Set dv = dx so v = x. |
| Two-sample t-test: equal variance assumed | Used n-1 denominator for pooled variance instead of (n1-1)+(n2-1) | Pooled variance: sum of (n-1)*s^2 for each group, divided by total df = n1+n2-2 |
| Eigenvalue calculation: 3x3 matrix | Expanded the 3x3 determinant along row 2 instead of row 1, made sign error on cofactors. | Expand along the row or column with the most zeros. Track cofactor signs with the checkerboard pattern. |
A completed error log entry captures the problem type, the precise failure point, and the correct method in the student's own words. Generic entries like 'made arithmetic error' are useless.
“Got the sign wrong” is a useless error log entry. “Forgot to distribute the negative when expanding -(3x - 2) in step 3” is a useful one. The goal is to describe the error precisely enough that when you return to this entry in three days, you know exactly what to watch for. Generic entries do not change behaviour; specific entries do.
Why Redoing Beats Reviewing Solutions
After logging an error, the natural instinct is to read the worked solution carefully until you understand it. That understanding is genuine but temporary. Reading the solution tells your brain what the correct path looks like. Redoing the problem from scratch the next day without looking at the solution forces your brain to reconstruct that path. The reconstruction effort is where the durable learning occurs.
This distinction maps directly to the testing effect documented by Roediger and Karpicke (2006): retrieval practice produces substantially better retention than restudying, even when the initial study session involves genuine understanding. In maths, “retrieval practice” means attempting the problem without reference to any solution or worked example. Students who read solutions but do not redo problems under retrieval conditions consistently underperform their own self-assessment suggests they should.
How to Use Spaced Practice on Maths Weak Spots
Spaced practice schedules the same material at increasing intervals, forcing retrieval just as the memory trace begins to fade. The effect compounds: each successful retrieval strengthens the memory further and extends the next optimal interval. For maths, this means returning to topics identified in your error log at expanding gaps rather than massing all your integration practice into one session.
Cepeda et al. (2006), reviewing 254 studies on the spacing effect, found that for material to be retained across periods of weeks or months, the study gap should expand from short intervals early to longer intervals later. A topic that went into your error log today should return tomorrow, then three days later, then a week later. By the third retrieval attempt, most students can execute the procedure without hesitation.
A Concrete Spaced Practice Schedule
Four weeks before your exam, the schedule runs roughly like this. Day 1, you identify and log errors from your first problem session. Day 2, you redo those specific problems. Day 5, you revisit the same topic with two or three new problems at the same difficulty. Day 12, you revisit with one new problem plus your hardest logged problem. Day 20, you complete a mixed set that includes this topic alongside others, which replicates exam conditions.
| Days before exam | Session focus | Session length |
|---|---|---|
| 28 days | Baseline problem set per topic: identify errors, start log | 60-90 min |
| 27 days | Redo all logged problems from scratch, no notes | 30-40 min |
| 24 days | Same topic: 3-4 new medium problems + hardest logged error | 40 min |
| 21 days | Different topic (rotates through all topics) | 60-90 min |
| 17 days | Mixed session: 2 problems from 3 different weak topics | 50 min |
| 10 days | Timed half-paper covering all topics | 90 min |
| 5 days | Full timed past paper under exam conditions | 2-3 hours |
| 1-2 days | Light review of error log summaries only; no new problems | 30 min |
A simplified four-week spaced schedule. The exact intervals adjust to your specific exam date; the principle is expanding gaps and mixing topics as the exam approaches.
Interleaving Topics Within a Session
Interleaving means switching between topics within a single study session rather than completing all problems from one topic before moving to the next. Rohrer and Taylor (2007) tested blocked versus interleaved practice on mathematics and found that students who practised interleaved achieved 43 percent higher scores on a delayed test, despite preferring the blocked approach during practice because it felt easier.
The mechanism is the same as for spaced practice: interleaving forces your brain to retrieve the correct method for each problem type rather than applying the most recent method automatically. Exams interleave topics by design. Revising with interleaved sessions trains exactly the discrimination skill the exam demands.
Interleaved practice feels harder and students typically rate their confidence as lower during interleaved sessions than blocked sessions. Yet delayed test scores are consistently higher. The difficulty is the mechanism: your brain works harder to retrieve the right method from scratch each time, which strengthens the retrieval path. If revision feels too easy, it is probably not building much.
How to Use Past Papers Under Timed Conditions
Past papers under timed conditions serve two purposes: they diagnose gaps while there is still time to close them, and they build the pacing fluency and pressure tolerance that problem sets alone cannot replicate. Most students use past papers too late and too gently. Both habits undermine the value.
The Timed Paper Protocol
Begin with full timed papers four weeks before your exam. Earlier feels uncomfortable because there will be topics you have not revised thoroughly, but that discomfort is precisely the point: you discover which topics matter most and how long each question type takes you to complete under real conditions. The discovery two weeks before the exam is not helpful. The same discovery four weeks before is.
Set up a genuine exam environment
Sit at a desk with only the materials your actual exam allows. Set a timer for the exact exam duration. No phone, no notes, no option to pause. Anything short of this understates your pacing risk.
Attempt every question
Do not skip questions you are uncertain about without at least reading them and writing the method you would attempt. Partial credit exists in most maths exams, and a method sketch that is directionally correct earns marks. Leaving a question blank earns nothing.
Mark with the official mark scheme
After completing the paper, mark it using the official or published mark scheme, not your own judgment. Your judgment will be more lenient than the examiner's. The official scheme shows which steps earn marks and which do not.
Log every error before moving on
Before closing the paper, enter every wrong answer into your error log with the exact error type and correct method. Do not move to the next revision task until the error log is updated. The paper without the log is half the value.
Marking and Error Analysis After Each Paper
The score you get on a past paper matters less than the error pattern it reveals. A student who scores 52 percent but identifies three specific recurring error types and targets them in the next fortnight will outperform a student who scores 65 percent, feels broadly satisfied, and moves on without systematic analysis.
After marking, categorise your errors into three types. Type A errors are careless arithmetic or procedural slips under time pressure: you knew the method but made a calculation mistake. Type B errors are conceptual gaps: you used the wrong method or could not identify the required approach. Type C errors are pacing failures: you ran out of time before completing a section you could have done correctly. Each type demands a different response. Type A responds to slowing down and checking work. Type B responds to targeted problem practice. Type C responds to timing drills on the specific question types where you ran over.
A Sample Weekly Maths Revision Plan
The structure below assumes roughly eight to ten hours of maths revision per week across a four-week period. Adjust the hours to your actual exam schedule and the number of topics your paper covers.
Weeks 3-4 before exam
- •Day 1-2: Build problem bank; first problem session per topic; start error log
- •Day 3: Redo all Day 1-2 error log problems from scratch
- •Day 4-5: Second topic; new problem session; update log
- •Day 6: First interleaved session (mix 2-3 topics)
- •Day 7: Rest or light review of log summaries only
Weeks 1-2 before exam
- •Day 1-3: Targeted problem sessions on weakest topics from log
- •Day 4: First full timed past paper, strict conditions
- •Day 5: Mark paper, update error log, targeted fixes
- •Day 6: Second full timed past paper
- •Day 7-8: Light review of error log; no new hard problems
Notice that the week immediately before the exam carries no new heavy learning. Deep conceptual gaps do not close in a week. That final week is for consolidating what you have already practised, running one more timed paper, and protecting your sleep and mental state. Students who try to learn new material in the final three days typically perform worse, not better, because the additional cognitive load disrupts retrieval of material they had already secured.
For help calculating your current grade and how much room your exam score leaves, the final grade calculator shows exactly what you need on the exam given your current standing and the exam's weight. The broader grade calculators hub covers weighted averages and module totals. For tools relevant to your specific maths course, the subject calculators hub collects the most commonly needed calculation tools in one place. If you need additional practice on specific problem types between exam sessions, the AI tutor can generate targeted problems and give immediate feedback.
Subject Calculators
Access calculation tools for your specific maths course, from calculus to statistics to linear algebra.
For the broader revision picture, the revision timetable guide walks through scheduling maths alongside other modules. If your exam is closer than four weeks, the one-week module revision plan provides a tighter timeline. Managing nerves on exam day itself is covered in the exam anxiety guide, and exam time management covers marks-per-minute pacing during the paper. The university resources hub collects study tools, calculators, and citation generators in one place.
Key Takeaways
- Maths revision builds procedural skill through problem-solving, not declarative knowledge through reading. The exam tests execution under time pressure; only practising execution trains it.
- Re-reading notes and reviewing worked solutions produce recognition familiarity, not retrieval fluency. Both feel productive. Neither prepares you to generate solutions under exam conditions.
- An error log (problem type, exact error, correct approach in your own words) is the highest-yield single tool in maths revision. Generic entries like "arithmetic error" are useless; specific entries that name the step and the correct method drive improvement.
- Redoing logged problems from scratch one to two days later, without any reference to solutions, builds durable procedural memory through retrieval practice. Reading the solution again does not.
- Spaced practice at expanding intervals (1 day, 3 days, 1 week) on weak topics compounds retrieval strength before the exam. Massing all practice on the same day degrades retention.
- Interleaving multiple topics within a single session outperforms blocked practice on delayed tests by roughly 40 percent (Rohrer and Taylor, 2007). It feels harder because it is harder, which is exactly why it works.
- Begin timed past papers under strict exam conditions at least four weeks before the exam. Categorise errors into Type A (slips), Type B (conceptual), and Type C (pacing), and apply the targeted fix for each category.


