Spaced Repetition Reminders: How Long to Leave Between Reviews

Almost every piece of study advice tells you to review things more than once. Very little of it tells you when — and the gap turns out to be the part that decides whether the second look was worth taking at all.

This is one of the better-evidenced findings in the psychology of learning, and it has a practical consequence: for the same total hours at the desk, changing only the spacing between sessions can move how much you still know months later. Not by a little.

How much difference does the gap actually make?

A large amount, and for free. In a study of over 1,350 people, researchers taught a set of facts, brought participants back for a single review after gaps of up to 3.5 months, then tested them up to a year later. Holding total study time fixed, the best gap produced a 64% increase in final recall compared with reviewing immediately.

That figure comes from Cepeda, Vul, Rohrer, Wixted and Pashler, "Spacing effects in learning: A temporal ridgeline of optimal retention", published in Psychological Science 19(11), 1095–1102 (2008). The effect size for recall was d = 1.1 — large, by the conventions of the field.

Read that sentence about total study time again, because it is the whole argument. The participants who did well did not work harder. They worked on a different schedule. The paper's own conclusion is blunt about what that implies: "The interaction of gap and test delay implies that many educational practices are likely to be highly inefficient."

So how long should the gap be?

Roughly 20% of the time until you need the material, for anything a few weeks out — dropping to about 5% when the thing you are aiming at is a year away. The optimal gap is not a fixed number of days. It scales with how far off the test is, which is why one-size study timetables misfire.

The paper reports the measured values directly. For target delays of 7, 35, 70 and 350 days, the gaps that produced the best recall were 1, 11, 21 and 21 days respectively.

Turned into arithmetic you can do in your head:

• Exam in a week? Review roughly a day later.
• Exam in five weeks? Leave it around ten days.
• Exam in ten weeks? Three weeks.
• Something you want to still know next year? Three weeks is still about right. The curve flattens; it does not keep stretching.

Notice what happens at the long end. Beyond a couple of months, the optimal gap stops growing much — 21 days serves both the 70-day and the 350-day target. If you are learning something you intend to keep, a review every three weeks or so is a defensible default rather than a guess.

Why does reviewing immediately feel better?

Because immediate review is easy, and ease feels like learning. Reading something twice in one sitting produces smooth, confident recall in that sitting. Come back after three weeks and the same material feels effortful and half-forgotten — which is uncomfortable, and is also the condition under which the review does its work.

This is the trap. The schedule that feels most productive at the desk is close to the one that performed worst a year later. Your sense of how well a session went is measuring fluency in the moment, not durability, and those two things come apart exactly where it matters.

There is no way to feel your way to the right answer here. It has to be scheduled in advance, by a version of you that is not currently in the room.

Where a reminder call fits, and where it doesn't

Use a reminder for the gap you would otherwise let slide — the one three weeks out, after the urgency has worn off. Flashcard apps already handle per-item intervals far better than any phone call could. What they do not do is get you to sit down at all on a Tuesday evening five weeks before an exam.

That is the honest division of labour. If you are drilling vocabulary or anatomy, use a spaced-repetition app; the algorithm is doing something a calendar cannot. But the failure mode for most people is not choosing the wrong interval for card 47. It is that week three arrives, nothing is due yet, and the whole plan quietly stops.

A scheduled call is useful precisely because it is hard to dismiss. In ReminderIt you can set one up by message — "remind me every 3 weeks on Sunday at 7pm to review my Spanish notes" — and it will read the schedule back before saving it. Set the gap once, when you are thinking clearly about the timetable, and let it arrive when you are not.

If you are working towards a fixed date, the better setup is two or three one-off reminders at the calculated gaps rather than a recurring one. Work backwards from the exam, apply the 20% rule, and put the calls in.

What this research does not show

It does not show that spacing makes you understand anything. The participants learned 32 obscure trivia facts with one-word answers — the example in the paper asks which European nation eats the most spicy Mexican food, and the answer is Norway. That is a memory task, cleanly measured, and deliberately so.

Whether a 20% gap is optimal for essay technique, mathematical problem-solving or a musical instrument is a genuinely open question, and anyone who tells you the number transfers unchanged is overselling it. What survives the caveats is the direction and the size: spreading a fixed amount of study out beats bunching it, by a margin large enough to care about, and reviewing too soon is a real cost rather than a neutral choice.

For anything that is substantially recall — vocabulary, terminology, dates, formulae, drug names, case citations — the evidence is about as good as it gets in this field. Put the gap in the diary and let it be uncomfortable when it comes round.