Spaced repetition: how it works, and how to start today

What is spaced repetition?

Spaced repetition is a way of reviewing. Instead of going over something many times in one sitting, you come back to it several times, with longer and longer gaps in between. Each review is a short test: you try to remember the answer, then check.

It combines two of the best-supported findings in learning research:

• Spacing. Reviews spread out over time beat the same reviews crammed together. This is the spacing effect.
• Retrieval. Pulling an answer out of memory strengthens it far more than reading it again. This is active recall, also called the testing effect.

The timing is what makes it efficient. Right after you learn something, you forget it quickly, then more slowly. That shape is the forgetting curve. A review just as an idea starts to slip takes a little effort and makes the memory last longer than before, so the next review can wait longer. After a handful of reviews spread over months, you keep the idea with a few minutes of total effort.

You can do it with paper cards in a box, with a calendar, or with an app that schedules each question for you. The rest of this guide covers all three.

The spacing effect: why spacing works

The first person to measure it was Hermann Ebbinghaus, the same psychologist who drew the forgetting curve. In his 1885 book on memory, he reported that for a list of 12 nonsense syllables, 38 repetitions spread over three days made relearning it on the fourth day just as easy as 68 repetitions in a row on the day before. Just over half the effort, spent at better moments.

The finding has been repeated in hundreds of experiments since. A 2006 review by Nicholas Cepeda and colleagues pooled 839 assessments of spacing from 317 experiments in 184 published articles. Its most useful conclusion: the gap that gives the best retention grows with how long you need to remember. In 2013, a review of ten popular study techniques led by John Dunlosky rated spaced practice one of only two with high utility. The other was practice testing.

How long should the gap be?

The clearest answer comes from a second study by the same team, published in 2008. More than 1,350 people learned 32 little-known facts (one example: "Who invented snow golf?"). They reviewed them once, after a gap of anything from zero to 105 days, then took a final test 7, 35, 70 or 350 days after the review. Here is the gap that worked best for each test delay:

Best of the gaps the study tested, for recall. Cepeda et al., 2008, Psychological Science.

Three lessons come out of it:

• The longer you need it, the longer the gap. As a share of the time until the test, the best gap fell from about 20 to 40% for a test a week away to about 5 to 10% for a test a year away.
• Too short is worse than too long. Recall rose steeply as the gap grew, then fell off only slowly past the best point. If you are unsure, wait a little longer rather than a little less.
• Spacing pays most for the long term. For a test a week away, spacing helped a little. For tests weeks or months away, it made a large difference.

How a schedule grows

Put the spacing effect and the forgetting curve together and you get the core idea of spaced repetition. Each successful review leaves a memory that fades more slowly than before, so the next review can wait longer. The gaps grow, and the number of reviews you need stays small. The chart at the top of this page shows the idea: five reviews, spread over six months, keep the curve high.

Five reviews cover about five months. The sixth would come about eight months later. Most of the work happens in the first few weeks, when the gaps are short and the memory is still fragile.

Do the gaps have to grow? Less than you might think. In lab studies, Jeffrey Karpicke and Henry Roediger found that equal gaps did as well as growing ones, or better, on a test two days later, and a later study by Karpicke and Althea Bauernschmidt found that the total amount of spacing mattered much more than whether the gaps grew, shrank or stayed the same. Growing gaps are still the practical choice when you want to keep something for months: they spread a few reviews over a long time.

Start today: a five-minute setup

1. Pick one thing you learned this week. A chapter, a lecture, a talk, a meeting. Something you would be annoyed to forget by next month.
2. Write three to five questions about it. On paper or in your notes app, with the answers out of sight. Ask for explanations and examples, not just names: "Why does spacing beat cramming?" is better than "Who discovered the spacing effect?"
3. Put the review dates in your calendar. Tomorrow, then 3 days later, a week later and 3 weeks later is a good start. The calculator does it in one click.
4. On each date, answer before you look. Say or write your answer, then check it. Recall first is what makes the review work.
5. Send misses back. Any question you got wrong comes back the next day, then follows the schedule again from there.

That is the whole method. Everything else, from index card boxes to algorithms, is a way of doing step 3 and step 5 for many questions at once.

The Leitner system: spaced repetition by hand

In the early 1970s, the German science journalist Sebastian Leitner described a way to run spaced repetition with nothing but flashcards and a box, in his book So lernt man lernen ("How to learn to learn"). The box is split into compartments. Every card starts in the first one. A card you answer correctly moves one compartment along, where it waits longer before you see it again. A card you get wrong goes back to the start.

Leitner's own box used compartments of different sizes, and you reviewed one when it filled up. Most people today use a fixed calendar instead. Here is a five-box version that is easy to run:

Number your days from 1 on a sticky note on the box. Every day, go through Box 1. On even days, also do Box 2. When the day number divides by 4, add Box 3; by 8, Box 4; by 16, Box 5. Each box waits twice as long as the one before, which gives you growing gaps without tracking a date for each card.

One card, step by step

Say you are reading Make It Stick and on day 1 you write this card:

Here is where it sits each day it comes up:

The gaps grow from 1 day to 2, 4 and 8. On day 16 you remember the "feels harder" part but not why it helps, so the card goes back to Box 1. It climbs again faster than the first time, because the later boxes come up on their own schedule, and by day 32 it sits in Box 5, where you see it every 16 days. Once a card has stayed in Box 5 for a few rounds, you can retire it or check it once a month.

The Leitner system works well for a few hundred cards. Its limits show with more: every card in a box waits the same time whether it was easy or hard, and a box of thousands of cards is a lot to carry around. That is the problem apps solve.

How apps schedule reviews: SM-2 and FSRS

Software can keep a separate gap for every card and adjust it after each answer. Two algorithms cover most of what you will meet. (Choosing an app? See our comparison of the best apps to remember what you read.)

SM-2: the classic

Piotr Wozniak created SM-2 in 1987 for SuperMemo, his spaced repetition program. Many apps still use a version of it, including Prismo. The rules fit in a few lines:

• After each review, you rate your answer from 0 (complete blackout) to 5 (perfect).
• The first gap is 1 day, the second is 6 days. After that, each gap is the previous one multiplied by the card's ease factor, which starts at 2.5.
• The ease factor goes up a little after a perfect answer (5), stays the same after a right answer with some hesitation (4), and goes down after a struggle (3) or a miss, but never below 1.3. Cards you find hard end up with shorter gaps.
• A rating below 3 counts as forgotten: the card starts again from a 1-day gap.

These are the gaps in the chart above. Six reviews take a card past a year. SM-2's strength is that it is simple and predictable. Its weakness is that it uses the same starting numbers for everyone and changes each card's gap only through that one ease factor.

FSRS: the newer scheduler in Anki

Anki started with its own variant of SM-2, with four answer buttons instead of six grades and a single "Again" button for a miss. Since version 23.10, it also ships FSRS, the Free Spaced Repetition Scheduler, developed by Jarrett Ye, a research engineer at the language-learning company MaiMemo, building on research he published in 2022. You turn it on in the deck options. FSRS models each card's memory with three numbers:

• Retrievability: the probability that you could recall the card right now.
• Stability: how many days it takes for that probability to fall from 100% to 90%.
• Difficulty: how hard the card is for you.

Instead of fixed multipliers, you choose a desired retention, the chance of remembering a card when it comes up (Anki's default is 90%), and FSRS schedules each review for when your predicted recall falls to that level. Its optimizer fits its parameters to your own review history. The FSRS developers estimate, from simulations, that it needs 20 to 30% fewer reviews than SM-2 for the same retention. That is their figure rather than an independent study, though the project publishes its code, benchmark and data openly.

Which algorithm you use matters less than whether you do the reviews. A simple schedule you follow beats a perfect one you abandon in week three.

Common mistakes

• Trusting how it feels. Cramming feels more effective because the material is fresh. In a 2009 flashcard study by Nate Kornell, spacing beat massed study for 90% of participants, yet after the first session 72% believed massing had worked better.
• Drilling small stacks. In the same study, one big stack of flashcards beat four small stacks studied one at a time. A big shuffled stack puts more time between each card's turns. Don't split your cards into small piles to drill until perfect.
• Rereading instead of recalling. Glancing at the answer and thinking "yes, I knew that" is recognition. Answer first, out loud or on paper, then check.
• Reviewing too soon. A review that feels effortless does little. Waiting until an idea has started to fade, while you can still bring it back, is the point.
• Adding too much at once. Every new card brings about five more reviews over the following months, and more if you miss it. Fifty new cards a day becomes well over a hundred reviews a day within a month. Add what you can keep up with.
• Spacing what you don't understand. Spaced repetition keeps whatever you put in. Memorizing a definition you can't explain gives you a definition you can't explain. Understand it first, then schedule it.
• Giving up after a missed week. A backlog is not a reset. Do the overdue reviews over a few days and carry on.

Flashcards for facts, questions for ideas

Spaced repetition grew up around flashcards: vocabulary, anatomy, formulas, capitals, dates. That is where it is easiest to use, because each fact fits on a card and the answer is either right or wrong.

Books, courses and talks are harder. Their value is in ideas, arguments and examples, and you rarely need to recite them word for word. The method still works, with a few changes:

• Write in your own words first. After a chapter, close the book and write what you remember. That first recall is already a review. Our guide on how to remember what you read has a template.
• Ask for explanations, not labels. "What does the author mean by X, and what is an example?" keeps the idea. "What is X called?" keeps a word.
• Fewer, bigger questions. Five questions on the main ideas of a book are worth more than fifty on its details, and you will actually keep up with them.
• Mix the formats. Explain an idea, apply it to a situation, compare it with another. Varied questions build a memory you can use, not one that only answers one prompt.

Want to see how much of a book you kept without any of this? Try the quiz on Make It Stick, written by Peter C. Brown with the memory researchers Henry Roediger and Mark McDaniel, or browse the other book quizzes.

Where Prismo fits

I built Prismo for the second kind of learning: ideas from books, courses, podcasts and conversations, where writing flashcards is the step most people skip. You write or dictate a few lines about what you learned. Prismo turns them into notes and quiz questions, then schedules them with an algorithm built on SM-2's ease factor: each right answer moves a question up a ladder of gaps, from a day to several months, stretched further when you answer quickly, and a wrong one moves it back down. You open it for a few minutes a day.

It is not the right tool for everything. If you want to write your own cards, share decks or tune every setting of the scheduler, Anki does that very well. And if you only need to keep one course for one exam, a calendar and the review schedule calculator are enough.

Sources

• Ebbinghaus, H. (1885). Über das Gedächtnis. English translation: Memory: A Contribution to Experimental Psychology, chapter 8 (section 34, the 68 and 38 repetitions).
• Cepeda, N. J., Pashler, H., Vul, E., Wixted, J. T., & Rohrer, D. (2006). Distributed practice in verbal recall tasks: A review and quantitative synthesis. Psychological Bulletin, 132(3), 354–380.
• Cepeda, N. J., Vul, E., Rohrer, D., Wixted, J. T., & Pashler, H. (2008). Spacing effects in learning: A temporal ridgeline of optimal retention. Psychological Science, 19(11), 1095–1102.
• Dunlosky, J., Rawson, K. A., Marsh, E. J., Nathan, M. J., & Willingham, D. T. (2013). Improving students' learning with effective learning techniques: Promising directions from cognitive and educational psychology. Psychological Science in the Public Interest, 14(1), 4–58.
• Karpicke, J. D., & Roediger, H. L. (2007). Expanding retrieval practice promotes short-term retention, but equally spaced retrieval enhances long-term retention. Journal of Experimental Psychology: Learning, Memory, and Cognition, 33(4), 704–719.
• Karpicke, J. D., & Bauernschmidt, A. (2011). Spaced retrieval: Absolute spacing enhances learning regardless of relative spacing. Journal of Experimental Psychology: Learning, Memory, and Cognition, 37(5), 1250–1257.
• Kornell, N. (2009). Optimising learning using flashcards: Spacing is more effective than cramming. Applied Cognitive Psychology, 23(9), 1297–1317.
• Leitner, S. (1972). So lernt man lernen. Freiburg: Herder.
• Wozniak, P. A. (1990). Optimization of learning: Algorithm SM-2. Master's thesis, University of Technology in Poznan.
• Anki Manual: Deck options, FSRS and What spaced repetition algorithm does Anki use?
• Open Spaced Repetition: ABC of FSRS and the SRS benchmark.
• Ye, J., Su, J., & Cao, Y. (2022). A stochastic shortest path algorithm for optimizing spaced repetition scheduling. Proceedings of the 28th ACM SIGKDD Conference on Knowledge Discovery and Data Mining, 4381–4390.