How to Revise for A-Levels Effectively

Most pages answering this question recommend the same three things — active recall, spaced repetition, past papers — and none of them says what the evidence is actually worth. This page does. It also gives the raw marks an A* took at AQA in June 2026, the assessment objective weightings that decide what your marks are for, and the June 2027 dates JCQ has already published.

What does the summer 2026 data say a good A-level grade actually is?

A revision plan needs a target, and the target should be a published number rather than a feeling. JCQ's summer 2026 press notice reports that 28.5% of A-level results across the UK were awarded A*–A, compared with 28.3% in 2025, and that 9.6% were awarded A*, compared with 9.4% in 2025. Those figures cover all five awarding bodies and all four UK nations. Read them the right way round: an A grade is roughly a top-quarter outcome and an A* is roughly a top-tenth one. The year-on-year movement is 0.2 percentage points in both cases — 28.5 minus 28.3, and 9.6 minus 9.4 — which is far too small to plan around. If someone tells you A-levels got easier this year, that is the size of the change they are describing.

The entry numbers move more than the grades do. JCQ counted 906,425 A-level entries at all ages in 2026 against 882,653 in 2025, a rise of 2.7%. The obvious inference is that competition is intensifying, and the obvious inference is wrong. The cover sheet of JCQ's subject-level table gives the reason: on ONS projections the 18-year-old population across England, Wales and Northern Ireland is estimated at 792,910, an increase of 3.9% on 2025. Entries rose 2.7%; the age group that sits them grew 3.9%. The cohort grew faster than the entry count did. Eighteen-year-olds supplied 830,769 of those entries, or 91.7% of the UK total, so the A-level population is overwhelmingly the standard sixth-form cohort rather than resitters and mature candidates.

Ofqual publishes an England-only set of figures that sits close to JCQ's but is not the same, and it is worth being careful about which one you are quoting. Across 835,950 A-level results in England, 28.3% were grade A or above and 78.1% were grade C or above. So 28.3% is England in 2026; 28.5% is the UK in 2026; and 28.3% is also JCQ's UK figure for 2025. Three different things, two of them sharing a number. If you see this statistic quoted without a country and a year attached to it, treat it as unreliable.

None of that tells you how to revise. It tells you what you are revising against, which is the part most revision guides skip. If you are aiming at an A*, you are aiming at a threshold fewer than one result in ten clears, and the marks that separate a strong A from an A* are almost never the recall marks. That is the argument the rest of this page makes, using the boards' own published weightings and boundaries rather than assertion. If you want structured help working through it, our A-level tuition pages set out how we work subject by subject.

Does the evidence behind active recall and spaced repetition hold up?

Almost every guide to A-level revision recommends the same short list: retrieval practice, spaced repetition, past papers and interleaving. The recommendations are broadly sound. What is missing almost everywhere is the size of the effect and the width of the error bars — and, awkwardly, the organisations that publish the effect sizes are the same ones publishing the caveats.

The Education Endowment Foundation's Teaching and Learning Toolkit rates metacognition and self-regulation approaches — planning your own study, monitoring whether it is working, changing it when it is not — at +8 months' additional progress. That is among the highest ratings in the Toolkit, and the same sentence adds that it can be "difficult to realise this impact in practice". The strand rests on 355 studies and the security of the evidence is rated high, but the EEF records that the topic lost a padlock because a large percentage of the studies were not independently evaluated. That is an unusually honest thing for a body to print about its own headline finding, and it sets the right frame for everything below: the direction of the evidence is clear, the magnitude in your bedroom in April is not.

Retrieval practice has the best reputation of the lot and a thinner base than the reputation suggests. The EEF's review of cognitive science approaches in the classroom compared retrieval with restudy across 21 studies. Only one of those 21 was delivered by regular class teachers; the rest were delivered by researchers, although all of them did take place in schools rather than in laboratories. On quizzing specifically, the review rates the evidence as moderate and generally positive when compared with no recap activity at all. Note what that comparison is against: doing nothing, not a well-designed alternative. The review is also open about the question that matters most at A-level, which is whether the benefit extends beyond rote factual recall into the applied and evaluative work that most A-level papers are largely made of.

Spacing has the same shape of problem. The review found 18 studies of spacing across days and lessons, and its own summary of the limitations is blunt: substantial variation between study results, with a significant number showing either no impact or a negative effect. For spacing inside a single lesson the evidence is thinner still. Both of the studies the review found were conducted in a single school in England, which the review says leaves insufficient variation in the population to test for generalisable impact.

The honest reading of all this is narrower than the usual one, and more useful. Retrieval and spacing are the best-supported things you can do with your own unsupervised time. The evidence for them was mostly generated by researchers rather than by students working alone, and nobody has demonstrated the size of the gain for an eighteen-year-old revising A-level Chemistry in April. Use them, and check them against something external — marked past-paper questions rather than a feeling of fluency. The same logic applies further down the school, which is why our GCSE tuition pages take the same approach for the same reason.

It is also why this page will not tell you how many hours a day to revise. The nearest thing to a published figure is the EEF's homework strand, which reports greater impact in secondary schools, +5 months, than in primary schools, +3 months — so independent study does appear to pay off more as students get older. But the security of that evidence is rated low, on 43 studies. A low-security +5 months is not a licence to convert hours into grades arithmetically. By contrast the feedback strand rests on 155 studies at a security rating of high, which is the best available reason to spend part of your revision time getting work marked by someone who knows the mark scheme rather than spending all of it alone.

Is interleaving proven at A-level, or only in maths?

An earlier version of this page recommended interleaving — mixing topics within a session rather than working through them in blocks — as a general A-level technique, alongside recall and spacing, with no qualification attached. That was more confident than the evidence supports, and it is worth saying so plainly rather than quietly deleting it.

The EEF review found 12 studies of interleaving. Its stated limitation is that 11 out of 12 of them were in maths, and that within those, many focused on interleaving tasks that required learners to select a solution strategy before implementing it. That is a precise finding about a precise skill: deciding which method a question calls for when the question does not tell you. It is genuinely valuable, and it is exactly what a mixed A-level Maths problem set trains.

What the review does not contain is evidence that the same manipulation helps someone revising History essay plans or English Literature set texts. It might. Nobody in that evidence base has shown it. If a revision guide tells you interleaving is proven to work across all your subjects, the reasonable question is which studies, in which subject, with which pupils.

Practically, that splits your subjects in two. In Maths and the quantitative parts of the sciences, mixed problem sets are well supported and easy to build: take a handful of past papers, cut them up, shuffle the questions and work them without knowing which chapter each one came from. In essay subjects the safer version of the same idea is not topic-shuffling for its own sake but answering unseen question stems on material you already know, which trains selection of argument rather than selection of method. Both are still retrieval practice; only one of them is interleaving as the studies defined it.

What are A-level marks actually awarded for?

"Revise the content" is the default instruction, and it is wrong in a specific and fixable way, because in most A-levels the content-recall objective is a minority of the marks. Every specification publishes its assessment objective weightings, and they differ enough between subjects that no single revision method can be correct for all of them.

Take AQA A-level Biology. AO1, knowledge and understanding, carries 30–35% of the marks overall. AO2, applying that knowledge, carries 40–45%, and AO3, analysis, interpretation and evaluation, carries 25–30%. Recall is the smallest of the three. The per-paper split in the same table is more instructive still: AO1 is 44–48% of Paper 1, 23–27% of Paper 2 and 28–32% of Paper 3. The same subject rewards flashcards close to twice as heavily on one paper as on another. Revise all three papers identically and you are over-preparing one and under-preparing another.

Pearson Edexcel A-level Mathematics runs the other way. Across the full GCE A level the weighting is 48–52% AO1, with AO2 and AO3 at 23–27% each. Roughly half the marks are for using standard mathematical technique accurately. Maths is the subject where drilling really does approach the whole job — and even there it leaves about half the paper for reasoning, modelling and interpretation, which drilling alone does not build.

AQA A-level History looks like the opposite case, and mostly is: AO1 carries 60%. But the remaining 40% splits evenly between AO2, the analysis of source material, and AO3, the historiographical strand, and those are skills rather than facts. The component weightings are 40%, 40% and 20%, with the Historical Investigation carrying that final fifth. AQA requires the Investigation to be extended writing "between 3500 and 4500 words in length, with a limit of 4500 words", which means a fifth of the qualification is coursework that cannot be revised for at all in the spring. It had to be done properly the first time.

AQA A-level English Literature A spreads its marks across five objectives: AO1 at 28%, AO2 and AO3 at 24% each, and AO4 and AO5 at 12% each. The non-exam assessment column of that table reads 5.6, 4.8, 4.8, 2.4 and 2.4, and 5.6 plus 4.8 plus 4.8 plus 2.4 plus 2.4 is 20.0 — so, as in History, a fifth of the qualification is banked or lost before study leave begins. The examinable portion is dominated by AO2 and AO3, which is to say by close reading and context rather than by plot recall.

The instruction that falls out of this is short. Open the specification for each of your subjects, find the scheme of assessment section, and write the AO weightings at the top of your revision list. Then check that the share of your time roughly matches the share of the marks. Most students find their revision is almost entirely AO1 in subjects where AO1 is a third of what is on offer. Our subject pages are organised the same way, by specification rather than by topic, because the specification is what the examiner is marking against.

What raw mark does an A* actually take?

"What percentage do I need for an A*" has no single answer, and the published boundaries show how wide the spread is. The table below gives AQA's raw mark boundaries from the June 2026 series for four of its largest A-levels.

Convert those and the point is hard to miss. An A* in A-level Mathematics took 260 of 300 marks, which is 86.7%. In Chemistry it took 255 of 300, or 85.0%. In Psychology, 229 of 288, or 79.5%. In Biology, 199 of 260, or 76.5%. Between Maths and Biology that is a gap of 10.2 percentage points on the same grade, in the same series, with the same board. A student targeting A*s in both is chasing two materially different standards, and a revision plan that treats them as one target is mis-allocating time.

The A boundaries tell the same story: 219 of 300 in Maths is 73.0%, while 174 of 260 in Biology is 66.9%. Look at the bottom of the table too, because it is rarely printed anywhere. An E in A-level Maths took 72 of 300, which is 24.0%, and an E in Biology took 63 of 260, or 24.2%. The pass threshold sits at roughly a quarter of the raw marks in both, which is why the distance between an E and an A* is far larger than students who talk about needing a few more marks tend to assume.

One caution matters more than any of the numbers above. AQA is explicit that once all exam papers have been marked, grade boundaries are set by senior examiners and assessment experts — for that series, on those papers. Last summer's boundary is a guide to the shape of a paper, not a threshold you can bank. If the June 2027 papers are harder, boundaries fall; if they are kinder, boundaries rise. Use them to compare subjects against each other, which is reasonably stable, rather than to decide how many marks you can afford to drop, which is not.

There is a related habit worth breaking. Marking your own past paper generously inflates every one of the percentages above. Mark to the published mark scheme, credit only what the scheme credits, and record the raw total against the boundary table rather than a rough impression of how it went. The gap between what a student thinks they scored and what the scheme actually awards is usually where the grade went.

How do you build a revision timetable from the published 2027 dates?

Most advice on this question is a version of "start three or four months before your June exams". The June 2027 series does not start in June. JCQ has already published the key dates, and the date of the first examination on the common timetable, covering GCSE and GCE alike, is 10 May 2027.

That single date changes the arithmetic of every plan built around June exams. Timetabling backwards from the start of June hands you several weeks that do not exist. Work backwards from 10 May 2027 instead, and treat everything after it as the tail of the series rather than the start of it.

Two other published dates belong in your calendar. The deadline for entries for the June 2027 GCSE and GCE series is 21 February 2027. That is a centre deadline rather than a student one, but if you are a private candidate or adding a subject late, it is the date that decides whether you sit at all. Results are released to candidates on 12 August 2027.

Check your own subject timetable rather than assuming last year's pattern holds, because JCQ moves papers between series. Following consultation on the June 2027 timetable, A-level Music Paper 1 was moved from the morning session of 11 June 2027 to the morning session of 9 June 2027. A two-day shift sounds trivial until it lands on top of another paper you had planned to revise for in that gap.

The method that follows is unglamorous and it works. List every specification topic for every subject — from the specification, not from your folder, because folders have gaps. Mark each topic red, amber or green against a past-paper question you have actually attempted and had marked, not against how familiar the words look on the page. Count the weeks you genuinely have between today and 10 May 2027, subtract the ones already spoken for by mocks, coursework deadlines and holidays you will not realistically work through, and distribute the red topics across what is left, with the assessment objective weightings from the previous section deciding how much time each subject gets. Then book a review of the plan for the week after you finish writing it, because the first version is always too optimistic. If you want it pressure-tested by someone who has seen the mark schemes, a consultation is the quickest way to get a second opinion on the sequencing.

How many A-levels should you take, and where do top grades cluster?

Ofqual's England data answers a question students ask constantly and guides rarely answer with figures: does taking a fourth A-level make top grades more likely? Among 18-year-olds in England in summer 2026, of the 14,975 candidates who took four A-levels, 1,900 — 12.7% — achieved A* in every one. Of the 193,275 who took three, 4,195 — 2.2% — achieved A* in all three.

12.7 divided by 2.2 is 5.8, so the clean-sweep rate is close to six times higher among four-subject candidates. It is tempting to read that as an argument for adding a fourth subject. It is not one. Nothing in that table says the fourth A-level caused the clean sweep, and schools generally permit a fourth subject to students already performing at the top of the cohort, so the four-subject group is a selected group before it starts. The absolute numbers make the point from the other side: 4,195 students swept A*s on three subjects against 1,900 on four. Far more clean sweeps come out of the three-subject route, because almost everyone takes that route.

For scale, 320,505 students in 2,835 schools and colleges in England received A-level results in summer 2026, alongside 1,047,750 students in 6,040 schools and colleges receiving GCSE results.

The centre-type and regional splits are where calibration gets uncomfortable. Across all 835,950 A-level results in England, 28.3% were grade A or above and 78.1% grade C or above. At grammar schools, on 83,585 results, 43.4% were grade A or above and 88.4% grade C or above. At further education establishments, on 51,280 results, the same two figures were 12.8% and 63.8%. Regionally, London's 168,505 results produced 32.7% at grade A or above, ahead of the South East on 31.5% from 166,940 results and the South West on 27.1% from 75,630.

Be careful what you conclude from that. A grammar school's 43.4% is not evidence that its teaching is several times better than a further education college's. Grammar schools select at entry, and FE colleges take a very different intake, including many resitters and part-time students. What the spread does tell you is that a national average is a poor benchmark for an individual. If you sit in a cohort where a third of results are grade A or above, your internal ranking predicts your outcome better than the national figure does; if you sit in one where the rate is far lower, the national figure is flattering you. Families weighing a move at the end of Year 11 will find the same selection effect discussed on our sixth form entry pages.

What should you change in your revision this week?

Four changes fall straight out of the numbers on this page, and none of them requires a new app, a new folder or a new highlighter.

First, print the assessment objective weightings for each of your subjects from the specification and check that your revision hours are distributed roughly the way the marks are. In AQA Biology, where AO1 is 30–35% of the total, a revision routine made almost entirely of flashcards is aimed at about a third of the paper.

Second, stop measuring progress in hours and start measuring it in marked marks. The EEF's homework evidence — +5 months at secondary, on 43 studies, at a security rating of low — is not strong enough to justify treating time as the unit of progress. The feedback evidence, 155 studies at a security rating of high, is strong enough to justify treating marked work as that unit instead.

Third, put 10 May 2027 at the top of the plan rather than the word June, and check your own board's timetable for papers that have moved.

Fourth, use last June's boundaries to compare subjects with one another rather than to set a personal mark target, because AQA sets each series' boundaries after that series has been marked.

And one thing not to do. Do not treat any of the effect sizes on this page as a promise. The most positive number here, +8 months for metacognition, arrives with the EEF's own warning that the impact is difficult to realise in practice, and the most-recommended technique of them all, retrieval practice, rests on a set of 21 studies in which exactly one was delivered by regular class teachers. The evidence is good enough to tell you which direction to walk in. It is not good enough to tell you how far you will get, and any page that hands you that number has made it up.

Frequently Asked Questions