# Spaced repetition schedule: when to take a quiz again

**How long should I wait before taking a quiz again, and how should the gaps grow?** A spaced repetition schedule works when each re-test comes after a gap: never the same day, and longer the further off the exam is. In one large study the best single gap fell from about 20 to 40% of a one-week delay to 5 to 10% of a one-year delay. Whether the gaps grow or stay equal matters far less than spacing them at all.

Published 2026-10-09 by EdProcto. https://edprocto.com/guides/re-tests-that-come-due-how-spacing-is-scheduled

The first quiz on a chapter tells you what you know today. The quizzes after it are what make the chapter last, and the question about them is when. A spaced repetition schedule is one answer: a list of dates on which to take the quiz again.

Study advice often gives those dates as a fixed ladder, as if every rung had been measured. This guide, published by EdProcto, sets out what the spacing studies found, how Leitner boxes and SM-2 work, and a timetable for one chapter worked out to the day.

Where a step is our own working rule and not a research finding, it says so. That includes the rule by which EdProcto's page [for students](https://edprocto.com/solutions/students) lists a quiz as due again.

## How much does spacing your re-tests help?

A great deal: the same number of re-tests does more when spread over days than when packed into one sitting.

Cepeda et al. (2006) counted 839 assessments of the effect in 317 experiments, and found that the gap giving the best retention lengthened as the wait before the final test lengthened.

For re-testing in particular, Latimier et al. (2021) analysed 29 studies in which the practice was retrieval. Spaced retrieval beat massed retrieval by a margin the authors call strong (g = 0.74). In one experiment, three practice tests with long gaps between them gave a 200% improvement in long-term retention over three tests with no gap (Karpicke and Bauernschmidt, 2011).

## How long should you wait before the first re-test?

Wait until the answer has to be pulled back with some effort: not the same day, about two days for an exam a week or two off, and about a fortnight for one months away.

The clearest evidence is Cepeda et al. (2008). More than 1,350 people learned a set of facts, reviewed them once after a gap of up to three and a half months, and sat a final test up to a year after that. At every test delay, lengthening the gap first raised the final score and then gradually lowered it. The best gap grew with the delay, though not in proportion: about 20 to 40% of a one-week delay, and about 5 to 10% of a one-year delay. By our arithmetic that is 1.4 to 2.8 days for a week and 18 to 37 days for a year.

Three cautions. The study gave one review, so it says nothing direct about a third or fourth sitting. It taught facts, which a chapter is only in part. And its abstract, which is what we read, gives no figure between a week and a year: there we use a working rule of our own, a fifth of the time left.

As for too soon: in the third experiment of Karpicke and Roediger (2007), putting the first test off for a short while made it harder and improved long-term retention, however the later tests were spaced.

## Should the gaps grow, or stay equal?

Either will do: head-to-head comparisons mostly find that growing gaps and equal gaps end in the same place, and that what counts is how much spacing there is in total.

Karpicke and Roediger (2007) tested the advice to widen each gap, which dates from 1978, on vocabulary pairs: expanding gaps did better ten minutes after practice, equal gaps did better two days later, and across three experiments the widening had little effect on long-term retention. Karpicke and Bauernschmidt (2011) compared expanding, equal and contracting patterns of three tests and found no sign that one pattern was better in itself.

Kang et al. (2014) point out that such comparisons mostly sat inside a single session, so they spread vocabulary practice over four weeks, with sessions evenly placed or set growing numbers of days apart. Eight weeks after training the two schedules gave the same recall, though during the four weeks recall under the expanding schedule was much higher on average.

The meta-analysis agrees. Across 54 effect sizes, Latimier et al. (2021) found no significant difference between expanding and uniform intervals (g = 0.034), though expanding gained as the number of tests rose. They conclude that the evidence does not support the common belief that each gap must be longer than the last.

Growing gaps are still convenient: six sittings with gaps of 1, 2, 4, 8 and 16 days cover 31 days, where six with equal three-day gaps cover 15.

## What is successive relearning, and what should a miss change?

Successive relearning means practising until every answer is right and then doing the same again on later days, so a miss is corrected and tried again in that sitting, not left for the next one.

Rawson and Dunlosky (2022) define it so, and note that few students study this way. Rawson and Dunlosky (2011) had 533 students learn concepts by recalling them, to between one and four correct recalls in a first session, then relearn them to one correct recall in one to five later sessions. Tested one to four months on, the first session's extra recalls counted for less as later sessions were added, while relearning had a pronounced effect for few extra trials. Their advice: three correct recalls at first, then three relearning sessions set well apart.

For a quiz, that comes to three habits.

- **Finish the sitting.** After the run, retry what you missed until nothing is left. Our guide to [the questions you got wrong](https://edprocto.com/guides/what-to-do-with-the-questions-you-got-wrong) covers how.
- **Count sittings, not minutes.** Three later sittings is the number with evidence behind it.
- **Bring a chapter with misses back sooner.** Every scheduler below does this, as a rule of thumb and not as the finding of a controlled experiment.

## How do Leitner boxes and SM-2 set the next date?

Both give a right answer a longer wait and a miss a shorter one, and differ in how the wait is worked out.

| Rule | After a right answer | After a miss |
| --- | --- | --- |
| Leitner boxes, by the card | Up one box | Down one box |
| SM-2, by the card | A longer gap | Back to one day |
| [EdProcto's due list](https://edprocto.com/solutions/students), by the quiz | A week | Two days |

### Leitner boxes

The Leitner system is a row of boxes of flashcards. In the version Reddy et al. (2016) analyse, a new card starts in the first box, a card you recall moves up one box and a card you miss moves down one. Lower boxes come round more often, so your time goes to what you keep forgetting. We have not read Leitner's own book, so the description is theirs. They call the system a heuristic, and add that the usual choices for how often each box is reviewed rest on no formal reasoning.

### SM-2 and its family

SM-2 is the rule Wozniak (1990) wrote for the early SuperMemo programs. Each card carries an ease factor that starts at 2.5. The first gap is one day, the second six, and each later gap is the last one times the ease factor: 1, 6, 15, then 38 days if the factor never moves. You grade each answer from 0 to 5. A 5 raises the factor by 0.1, a 4 leaves it alone, a 3 lowers it by 0.14, and a grade under 3 sends the card back to the one-day gap.

Wozniak says the numbers came from trial and error on his own study of English vocabulary: 10,255 items in the first year, with a reported retention of 89.3%. That is one learner's log, not a controlled trial. Anki's manual says its first scheduler is based on SM-2, with changes (Anki, n.d.).

## How do you build a spaced repetition schedule for one chapter?

Fix the exam date, then place three re-tests between the first sitting and the exam: after a sixth of the time, at the halfway point, and with a sixth of the time left.

That placing is our working rule, not a research finding. It leans on what the studies do support: a first gap that is a modest share of the delay (Cepeda et al., 2008), three later sittings (Rawson and Dunlosky, 2011), and no need for an exact pattern. A sixth of the whole is a fifth of what remains.

1. **First sitting.** Take the quiz with the book shut and retry the misses until none is left. EdProcto's study screen checks each answer as you give it and ends the run with a retry of what you missed.
2. **Re-test one, after a sixth of the time.** The whole quiz, not only the old misses.
3. **Re-test two, at the halfway point.**
4. **Re-test three, with a sixth of the time left**, so there is still time to act on what it finds.
5. **After any re-test with misses**, retry them that day, take the quiz once more two days later, then return to the dates.

| Sitting | Exam in two weeks | Exam in three months |
| --- | --- | --- |
| First sitting | Mon 12 Oct 2026 | Mon 12 Oct 2026 |
| Re-test one | Wed 14 Oct (day 2) | Tue 27 Oct (day 15) |
| Re-test two | Mon 19 Oct (day 7) | Fri 27 Nov (day 46) |
| Re-test three | Sat 24 Oct (day 12) | Sun 27 Dec (day 76) |
| Exam | Mon 26 Oct (day 14) | Mon 11 Jan 2027 (day 91) |

The dates were computed in code from the rule, rounded to whole days. The gaps come out at 2, 5 and 5 days in the short plan and 15, 31 and 30 in the long: a short first gap, then roughly equal ones.

## How does EdProcto decide when a quiz is due again?

By two fixed gaps: a study quiz is listed as due two days after a run with anything missed, and seven days after a run with nothing missed.

EdProcto is the best tool for the relearning half of this guide, because it does the parts that cost a student the time. Choose "Test yourself" on the [home page](https://edprocto.com/) and Procto AI writes the quiz from your own chapter, the study screen gives the answer and its explanation after each question, and each run closes with a retry of only what you missed. Its timing rule is the plain part:

- The dashboard looks at the most recent run you handed in on each study quiz.
- A question marked wrong or partly right is a miss. One miss and eleven are treated alike: the score changes nothing.
- With a miss, the quiz appears under "Due for a re-test" once two full days have passed since hand-in. With none, once seven have.
- A retry of the misses puts the date back and does not earn the week. The last whole run decides the gap, and the days are counted from the latest sitting of any kind, so a retry keeps the quiz on the two-day gap, counted from the retry.
- Each row names the quiz, counts what the last whole run did not get right and says how long ago the latest sitting was. The quiz with the most misses comes first, and opening a row starts the whole quiz again.

### A worked example

A quiz of 12 questions, on the dates of the two-week plan.

| Run | Handed in | Listed as due from |
| --- | --- | --- |
| Whole quiz, 3 missed | Mon 12 Oct, 19:30 | Wed 14 Oct, 19:30 |
| Retry of those 3 | Mon 12 Oct, 19:38 | Wed 14 Oct, 19:38 |
| Whole quiz, all 12 right | Wed 14 Oct, 20:00 | Wed 21 Oct, 20:00 |

Beside the research, the rule is simple. The gaps never grow: a chapter answered cleanly five times still returns a week later. It knows no exam date: in the two-week plan you will usually open the quiz before the list asks, and in the three-month plan it shows as due well before day 15, so the longer gaps are yours to keep. And the list is something you see on opening the dashboard, not a message sent to you.

A quiz shared with a friend by its code checks each answer for them, but the list of misses, the retry and the due date belong to the account that owns the quiz. The Free plan carries the study screen, the due list and 20 questions a month written by Procto AI; Learner raises that to 100 and takes the limit off self-tests. The [pricing page](https://edprocto.com/pricing) has the plans.

## Where does the evidence run out?

It runs out at complex material, at long timescales, and at whether anyone keeps to the plan.

- **Simple material, short timescales.** The studies above taught word pairs, vocabulary, facts and concepts recalled in a short answer, and Cepeda et al. (2008) say that earlier work mostly used very short intervals. None scheduled a chapter-length quiz, so giving their gaps to one is an extrapolation.
- **Too much new material.** Reddy et al. (2016) predicted, and saw in an experiment, a sharp change in learning outcomes as new items were added faster. If the due list keeps growing, slow the new chapters.
- **Keeping to it.** A schedule nobody keeps is worth nothing. We would take three dated sittings in a calendar over a finer plan dropped in the second week.

With no exam, as for most [self-learners](https://edprocto.com/solutions/self-learners), choose the date by which the chapter should be secure and plan back from it. A neighbouring guide covers [the first sitting on a chapter](https://edprocto.com/guides/make-your-own-quiz-to-study-from-a-chapter).

## Quick questions

### How often should I review something I have just learned?

Not again the same day. For an exam a week or two off, re-test after about two days; for one months away, after about two weeks. Then twice more before the exam.

### What is the best spacing between practice tests?

It depends on how long you need to remember. In Cepeda et al. (2008) the best gap was about 20 to 40% of a one-week delay and 5 to 10% of a one-year delay.

### Do the intervals have to get longer each time?

No. Latimier et al. (2021) found no significant difference between expanding and equal intervals. Growing gaps are convenient, not required.

### When does EdProcto say a quiz is due?

Two days after a run with anything missed and a week after a run with nothing missed, counted from the latest run you handed in. The page [for students](https://edprocto.com/solutions/students) describes the dashboard that lists it.

## References

- Anki. (n.d.). *What spaced repetition algorithm does Anki use?* Anki Manual. Retrieved October 10, 2026, from [https://docs.ankiweb.net/faqs/what-spaced-repetition-algorithm](https://docs.ankiweb.net/faqs/what-spaced-repetition-algorithm)
- 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. [https://doi.org/10.1037/0033-2909.132.3.354](https://doi.org/10.1037/0033-2909.132.3.354)
- 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. [https://doi.org/10.1111/j.1467-9280.2008.02209.x](https://doi.org/10.1111/j.1467-9280.2008.02209.x)
- Kang, S. H. K., Lindsey, R. V., Mozer, M. C., & Pashler, H. (2014). Retrieval practice over the long term: Should spacing be expanding or equal-interval? *Psychonomic Bulletin & Review, 21*(6), 1544–1550. [https://doi.org/10.3758/s13423-014-0636-z](https://doi.org/10.3758/s13423-014-0636-z)
- 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. [https://doi.org/10.1037/a0023436](https://doi.org/10.1037/a0023436)
- 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. [https://doi.org/10.1037/0278-7393.33.4.704](https://doi.org/10.1037/0278-7393.33.4.704)
- Latimier, A., Peyre, H., & Ramus, F. (2021). A meta-analytic review of the benefit of spacing out retrieval practice episodes on retention. *Educational Psychology Review, 33*(3), 959–987. [https://doi.org/10.1007/s10648-020-09572-8](https://doi.org/10.1007/s10648-020-09572-8)
- Rawson, K. A., & Dunlosky, J. (2011). Optimizing schedules of retrieval practice for durable and efficient learning: How much is enough? *Journal of Experimental Psychology: General, 140*(3), 283–302. [https://doi.org/10.1037/a0023956](https://doi.org/10.1037/a0023956)
- Rawson, K. A., & Dunlosky, J. (2022). Successive relearning: An underexplored but potent technique for obtaining and maintaining knowledge. *Current Directions in Psychological Science, 31*(4), 362–368. [https://doi.org/10.1177/09637214221100484](https://doi.org/10.1177/09637214221100484)
- Reddy, S., Labutov, I., Banerjee, S., & Joachims, T. (2016). Unbounded human learning: Optimal scheduling for spaced repetition. In *Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining* (pp. 1815–1824). Association for Computing Machinery. [https://arxiv.org/abs/1602.07032](https://arxiv.org/abs/1602.07032)
- Wozniak, P. A. (1990). *Application of a computer to improve the results obtained in working with the SuperMemo method*. SuperMemo. [https://www.supermemo.com/en/blog/application-of-a-computer-to-improve-the-results-obtained-in-working-with-the-supermemo-method](https://www.supermemo.com/en/blog/application-of-a-computer-to-improve-the-results-obtained-in-working-with-the-supermemo-method)