How spaced repetition actually works
The gap between reviews isn't wasted time — it's the ingredient. What the spacing effect is, how to get most of the benefit with no software at all, and the situations where it's the wrong tool.
The short version
- The same total study time, split across days, produces far more durable memory than the same time in one block.
- Partial forgetting between reviews is what makes the next review effective. Reviewing too soon wastes it.
- Longer gaps generally help more, up to the point where you've forgotten so much that the review becomes re-learning.
- You do not need scheduling software. A rough expanding sequence — a day, three days, a week, a fortnight — captures most of the benefit.
Almost everyone has run the experiment: two subjects, both revised for six hours. One got six hours the night before. The other got one hour a week for six weeks. Everyone knows how that comes out three months later, and yet the night before is still the default plan, because the deadline is what forces the work.
The spacing effect is unusually well established — the first systematic study of it is over a century old, and a meta-analysis covering hundreds of experiments found the same pattern across ages, materials and delays. It is also unusually cheap: it doesn't ask you to work harder or longer, only to distribute what you were going to do anyway.
Why gaps help
The uncomfortable version of the explanation is that forgetting is the mechanism. When you review something immediately, it's still sitting in working memory and the "review" costs nothing — no retrieval happens, so nothing is strengthened. Leave it a few days and it has become genuinely hard to reach. Reaching for it anyway is where the work is, and the work is what produces durability.
This is the same principle behind retrieval practice, and the two are not really separate techniques — spacing is what makes retrieval difficult enough to be worth doing. Spaced re-reading helps a little. Spaced self-testing is where the large effects are.
There's a second mechanism worth knowing about. Each time you study something you encode it alongside incidental context: where you were, what you'd just been thinking about, what mood you were in. Study on six different days and the memory acquires six different sets of cues. That makes it reachable from more directions later — including from the unfamiliar direction of an exam hall.
The counter-intuitive bit
A review where you struggle and eventually get there does more for you than one where the answer arrives instantly. If a review felt easy, the gap before it was too short. Ease is a signal to space further out, not a sign you've finished.
How long should the gaps be?
The honest answer is that the research gives a direction rather than a formula. The clearest finding is that the optimal gap scales with how long you need to remember the material: if the exam is a month away, gaps of days are right; if it's six months away, gaps of weeks are. As a rough guide, spacing reviews at somewhere around a tenth to a fifth of your target retention interval is a defensible starting point — and being roughly right matters far more than being precisely right.
The second finding is that gaps should expand. Early on, a memory is fragile and needs a review within a day or so. Once it's held for a week, it can hold for two. A workable default:
| Review | When | What it's for |
|---|---|---|
| 1st | Same day, a few hours later | Catch what never encoded at all |
| 2nd | Next day | The steepest part of the forgetting curve |
| 3rd | 3–4 days later | First genuinely effortful retrieval |
| 4th | 1–2 weeks later | Consolidation |
| 5th | 1 month later | Maintenance |
Adjust by difficulty, not by schedule. Anything you got wrong drops back to a short gap; anything that came instantly moves to the next one out. That single rule is most of what scheduling software does.
Doing it without software
Spaced-repetition apps compute a per-item interval from your grading history. They're good at that, and for large fact banks — vocabulary, anatomy, pharmacology — they're worth the setup. For a school or university term, the overhead often exceeds the benefit, and there are two much simpler systems that work.
The date list
Keep one page per subject. Every time you finish a topic, write the topic and the date. Once a week, go down the list and test yourself on anything last touched more than a week ago — oldest first. That's it. It's crude, it takes about a minute to maintain, and it captures the large majority of the spacing benefit because it guarantees that nothing sits untouched for a month by accident.
Three boxes
The physical version, and it's over sixty years old. Split your cards or questions into three piles: shaky, getting there, solid. Review shaky every session, getting-there every few sessions, solid every couple of weeks. Get one right, move it up a box; get it wrong, it goes straight back to shaky. The schedule emerges from the sorting.
Spacing needs a memory of what you covered when. YourSubjects tracks which sections you've read, which cards you've marked learned, and every practice and paper attempt with its date — so "what haven't I touched in three weeks?" is answerable without keeping a separate list.
Open the appWhere spacing goes wrong
Reviewing without retrieving. Spacing your re-readings is a weak version of the technique. If a review consists of looking at the answer, the gap did nothing for you. Cover the answer first, every time.
Gaps so long the material is gone. There's a difference between struggling to recall and having nothing to struggle with. If a review is 90% blank, you're re-learning at full cost, and the schedule needs tightening.
Spacing before you understand. Spaced repetition maintains knowledge; it doesn't create it. Distributing your reviews of a topic you never followed just distributes the confusion. Get to comprehension first, then schedule.
Reviewing everything at a uniform rate. The point of any spacing system is that easy material drops out of rotation and hard material keeps coming back. A system that treats every item the same is just a rota, and it will spend most of your time on things you already know.
What about cramming?
Cramming is not useless — it's genuinely effective for a test tomorrow. Massed study puts a lot into memory quickly, and if the only requirement is that it survive eighteen hours, it will. What it doesn't do is last, and it doesn't build anything the next module can stand on.
The realistic position: if the exam is tomorrow, cram, and don't feel bad about it. If it's more than a fortnight away, spacing is strictly better for the same hours. And if the subject is one where next year's material assumes this year's, cramming is borrowing at a rate you'll notice.
If you take one thing
Don't finish with a topic. Finish with a topic for now, note the date, and put it back in the rotation. The habit that matters isn't the interval arithmetic — it's never letting anything drop out of circulation entirely.
Where this comes from
- 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).
- 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). The source of the "optimal gap scales with retention interval" result.
- Ebbinghaus, H. (1885). Über das Gedächtnis — the original forgetting-curve work.
- Dunlosky, J., et al. (2013). Improving students' learning with effective learning techniques. Psychological Science in the Public Interest, 14(1). Rates distributed practice "high utility".
- Leitner, S. (1972). So lernt man lernen — the box system described above.