Does spaced repetition work for maths?

Yes, with one adjustment: what needs spacing in maths is problem-solving, not recall. Re-deriving and re-solving on a schedule keeps skills alive the way rereading never does. The evidence behind spacing and testing is some of the most robust in learning science; the hard part has always been running the system, not the science.

Can I use Anki for maths?

Many people do, and the ones it works for tend to be highly systematic. The costs are real though: you author every card yourself before any learning happens, cards drift toward memorising steps rather than solving, and a missed week becomes a review avalanche. If you love the craft of card-making, Anki rewards it. If you want the retention without the second job, you want the schedule built into the learning itself.

How often should I review maths topics?

On an expanding schedule: soon after learning, then at growing intervals, each review timed to land just before you'd forget. The right intervals differ per person and per topic, which is exactly why this belongs to software rather than to a calendar and willpower.

Why do I forget maths so fast even when I understood it?

Because understanding creates a memory, and untouched memories decay; most of what anyone learns once is gone within weeks. Forgetting is not a personal failing, it is the default. Retention is a schedule, not a talent.

Method

Spaced repetition for maths, without making a single card

You learned it once. Nothing was keeping it. Spaced repetition is the keeping part, and for maths it can work without the flashcard factory.

Updated September 2026

The science is settled. The chore is the problem.

Spacing reviews out over expanding intervals, and testing yourself rather than rereading, are about as close to consensus as learning science gets. The communities that live by this, medical students and language learners above all, retain volumes of material that look superhuman from outside. And yet most people who try spaced repetition for maths quit within months. Not because the method failed. Because the method, as usually practised, is a second job.

The standard tooling asks you to write every card yourself before any learning happens, to decide what is card-worthy while you are still a novice in the topic, and to show up every single day, because the schedule does not care that you had a bad week. Miss one and the backlog compounds into the avalanche every Anki user eventually posts about. The people who thrive under that regime are real, and they are not most people.

Maths changes what a “review” should be

There is a deeper mismatch too. Flashcards excel at facts: words, dates, doses. Maths is only partly facts. The thing worth keeping is the ability to solve, and a card that shows you the formula until you recognise it teaches recognition, not solving. Reviewing maths properly means being handed a fresh problem at the right difficulty, working it, and being told honestly how it went. That is hard to hand-author into cards, which is why so many maths decks quietly rot into step-memorisation.

How long the gap should actually be

The interval is not a matter of taste, and the research is unusually specific about it. Cepeda, Vul, Rohrer, Wixted and Pashler taught more than 1,350 people a set of facts, left a gap of up to three and a half months, gave them a single review, then tested them at delays of up to a year. The best gap turned out to depend on how long you need to keep the material: roughly 20 to 40 per cent of a one-week horizon, falling to roughly 5 to 10 per cent of a one-year horizon. Read that backwards and it becomes a planning rule. If you want a topic to survive until next week, review it in a day or two. If you want it next year, reviews spaced a few weeks to a month apart are the right neighbourhood, and going back every Sunday is effort you will feel as boredom long before you feel it as retention.

That is also the real argument against running the schedule off a calendar. One fixed interval cannot serve a topic you met yesterday and a topic you have been solid on since March, and nobody tracks a hundred separate clocks by hand. The schedule has to be per concept, and it has to move in response to how the last attempt actually went.

A schedule you can run by hand this week

If you are not using software yet, here is the version that fits on a sticky note. Day one: learn the topic and solve three problems. Day two: solve three fresh problems on the same topic without opening your notes. Day four: three more. Day eight: three more. From there, roughly double the gap every time the review goes cleanly, and drop it back to two days the first time it does not. The exact numbers are not sacred; the doubling is. The expanding gap is what does the work, and the fresh problems, rather than a reread, are what make it maths practice rather than trivia.

Two rules keep the hand-run version alive. Review the way you will be tested: on paper, from a blank page, with the working written out, because for a procedure the writing is the retrieval. And keep formulas and methods on separate clocks. A formula is a fact and a card will hold it; a method is a skill, and only solving keeps it.

What it looks like when the system carries itself

Picture the same science with the labour removed. The content already exists, connected on a map of every concept and what it stands on, so there is nothing to author. Each concept you learn enters its own review schedule automatically, and reviews arrive as newly generated problems, so you are always solving, never reciting. Because the map knows what depends on what, reviewing one concept quietly refreshes the ones underneath it. And when life eats a week, the schedule adapts instead of piling up a debt of six hundred cards and a side order of guilt.

The honest comparison with the flashcard route is not that one works and the other does not. It is about where the effort goes: into maintaining the system, or into the maths. If your attention is a scarce resource, that difference is the whole game.

When spacing is the wrong tool

Two cases where none of the above applies. The first is an exam in five days. At that horizon there is no expanding schedule to build, only the ordinary advice to spread those five days out instead of spending Saturday on all of it. The second matters more. Spaced repetition preserves; it does not teach. If a concept never worked in the first place, usually because the thing underneath it is missing, review will hand you the same failure at growing intervals and call it a schedule. Retention software is very good at keeping what you have and completely blind to what you never had.

Retention is the second problem, not the first

Which is why the schedule is not where to begin. Retention only matters for things worth retaining, so the first step is not a review schedule; it is knowing what you should be learning at all. If you are returning to maths after years away, start with where you actually are, learn forward from there, and let the keeping happen on schedule, in the background, indefinitely.

Questions

Does spaced repetition work for maths?: Yes, with one adjustment: what needs spacing in maths is problem-solving, not recall. Re-deriving and re-solving on a schedule keeps skills alive the way rereading never does. The evidence behind spacing and testing is some of the most robust in learning science; the hard part has always been running the system, not the science.
Can I use Anki for maths?: Many people do, and the ones it works for tend to be highly systematic. The costs are real though: you author every card yourself before any learning happens, cards drift toward memorising steps rather than solving, and a missed week becomes a review avalanche. If you love the craft of card-making, Anki rewards it. If you want the retention without the second job, you want the schedule built into the learning itself.
How often should I review maths topics?: On an expanding schedule: soon after learning, then at growing intervals, each review timed to land just before you'd forget. The right intervals differ per person and per topic, which is exactly why this belongs to software rather than to a calendar and willpower.
Why do I forget maths so fast even when I understood it?: Because understanding creates a memory, and untouched memories decay; most of what anyone learns once is gone within weeks. Forgetting is not a personal failing, it is the default. Retention is a schedule, not a talent.