How to use spaced repetition properly
Spacing your reviews out is one of the few study techniques with genuinely strong evidence behind it. It is also one of the easiest to do badly. Here is the interval that works, what the research actually says, and the three mistakes that stop people getting the benefit.
The interval depends on your exam date
There is no universally correct number of days, which is why generic advice about “review after 1 day, 3 days, 7 days” is only sometimes right. Cepeda and colleagues (2008) found the optimal gap scales with how far away the test is — approximately 10–20% of the time until you are examined.
| Test is in… | Roughly review every… |
|---|---|
| 1 week | 1 day |
| 1 month | 3–6 days |
| 3 months | 1.5–2.5 weeks |
| 6 months | 3–4 weeks |
The practical implication is counter-intuitive: cramming the night before is bad, but so is reviewing something daily when the exam is half a year away.
How good is the evidence?
Better than almost anything else in study advice. Cepeda et al. (2006) reviewed 317 experiments across 184 articles. Of 271 direct comparisons, only 12 failed to find a benefit from spacing.
Dunlosky et al. (2013) assessed ten widely used study techniques and rated exactly two as high utility: distributed practice and practice testing. Highlighting and rereading — the two things most students actually do — came out low.
Mistake 1: reviewing instead of retrieving
Turning a card over and thinking “yes, I knew that” is recognition, not recall, and it feels almost identical while being worth far less. Say the answer out loud or write it before you flip.
Adesope, Trevisan and Sundararajan (2017), a meta-analysis of 188 experiments, put practice testing ahead of rereading at an effect size of g = 0.51. Solid and real, rather than miraculous. That is worth saying, because the numbers usually quoted are the outliers.
The eye-catching one is Karpicke and Roediger (2008): about 80% recall a week later against 36% for repeated study. That was foreign-vocabulary learning under specific conditions. Their 2006 study on prose passages, which is much closer to revising a textbook chapter, found 56% against 42%. Both are true. Neither is 80% for everything.
Mistake 2: a deck you cannot finish
The most common way spaced repetition fails is a backlog. Two hundred cards due, you do sixty, tomorrow it is two hundred and forty, and by Thursday you have stopped opening the app.
If the queue is regularly going unfinished, the deck is too big. That is a sizing problem, not a discipline problem, and the fix is to cut it.
Mistake 3: spacing the wrong things
Time spent on cards you already know is close to wasted. What should dominate the schedule is what you have demonstrably got wrong, not what felt shaky, which is a poor guide.
In Laxu AI this is automatic: questions you miss in a quiz become flashcards and enter the schedule, so the material you actually failed on is prioritised over the material you merely felt unsure about.
Common questions
What is the optimal interval between reviews?
Cepeda and colleagues (2008) found the optimal gap scales with how far away the test is: roughly 10–20% of the time until you are examined. For a test in 10 days that is a gap of about 1–2 days; for a test in 6 months it is closer to 3–4 weeks. There is no single correct number of days: the right interval depends on when you need to know it.
How strong is the evidence for spaced repetition?
Unusually strong for education research. Cepeda et al. (2006) reviewed 317 experiments across 184 articles; of 271 direct comparisons, only 12 failed to show a benefit from spacing. Dunlosky et al. (2013) assessed ten widely used study techniques and rated only two as high utility: distributed practice and practice testing. Highlighting and rereading, the two things most students actually do, were rated low.
Is spaced repetition the same as active recall?
No, they are two different things that work best together. Active recall is retrieving the answer from memory instead of rereading it. Spacing is when you do that retrieval. Adesope et al. (2017), a meta-analysis of 188 experiments, found practice testing beat rereading with an effect size of g = 0.51. Spacing those tests out is what compounds the effect.
Where does the 80% retention figure come from?
Karpicke and Roediger (2008), published in Science, where repeated retrieval produced about 80% recall a week later against 36% for repeated study. That was foreign-vocabulary learning under fairly specific conditions. Their earlier 2006 study on prose passages, which is closer to revising a textbook chapter, found 56% against 42%. Both are real; be wary of any study app quoting the bigger number without saying which paper it came from.
How many cards should I review a day?
Fewer than you think, consistently, beats a large number occasionally. The failure mode is building a backlog you then avoid. If a daily queue is regularly going unfinished, the deck is too big rather than you being undisciplined. Cut it.
Does Laxu AI schedule reviews automatically?
Yes. Flashcards generated from your material go into a spaced repetition schedule, and questions you get wrong in a quiz become cards automatically, so gaps you actually demonstrated get prioritised over ones you guessed at.