Spaced repetition: how to memorise better over several weeks

Spaced repetition is particularly effective when, at each review, the student first tries to retrieve the information without looking at their notes. Research on distributed practice shows a solid advantage of spacing for long-term memorisation.

Rereading a revision sheet four times on the same evening is therefore not spaced repetition. Conversely, waiting a month before coming back to a concept that has been almost entirely forgotten is not automatically a good strategy. The aim is to spread reviews over time and use each one to check what is genuinely still available.

1. The principle of spaced repetition

Massed practice concentrates several repetitions in a short period. Distributed practice spreads a comparable amount of work over several moments.

A meta-analysis by Cepeda and colleagues synthesised 839 assessments from 317 experiments on distributed practice. The results confirm the effect of spacing on retention and show, above all, that the relevant interval between two study episodes depends on how long the information will need to be retained.

In other words, a student preparing for a test in five days and another who wants to retain vocabulary for six months should not necessarily use the same schedule.

The synthesis by Dunlosky and colleagues also ranks distributed practice, along with practice testing, among the techniques with the highest utility rating out of the ten strategies studied.

Spacing does not necessarily mean systematically increasing the intervals

Spaced repetition is often presented as a sequence such as D1, D3, D7, D15, D30. Increasing intervals can be a practical way to organise things, but they are not a universal scientific rule.

A meta-analysis by Latimier, Peyre and Ramus covering 29 studies found a large advantage of spaced recall over massed recall. However, it found no significant overall advantage of a schedule with gradually increasing intervals over a schedule with regular intervals.

The best-established principle is therefore simple: leave a genuine delay between several reviews. The exact schedule then needs to be adapted.

2. Why forgetting is part of learning

The aim, obviously, is not to forget. But information you have just read is still very accessible. Repeating it immediately can give an impression of mastery without really testing whether it will remain available over time.

After a delay, retrieving the same information takes more effort. This difficulty can be useful if the knowledge is still retrievable. Research on “desirable difficulties” describes precisely those learning conditions that seem less easy in the short term but do more to support later retention.

The review by Carpenter, Pan and Butler published in Nature Reviews Psychology synthesises several decades of research on spacing and retrieval practice. The authors also stress that these strategies remain underused, notably because they can feel less fluent than methods such as rereading.

The right signal: having to search, not starting from scratch

A spaced review becomes particularly informative when the answer no longer comes immediately but can still be reconstructed. For example:

• the definition comes back after a few seconds;
• the formula is recalled but a condition for applying it is missing;
• the steps of a method are known but their order is uncertain;
• a vocabulary word is recognised but hard to produce.

These hesitations show what needs strengthening. If the student can no longer recall the concept at all, you need to go back briefly to the explanation or the material, correct, then try again later. Too much difficulty is not automatically beneficial.

3. How to choose review intervals

The schedule should first depend on the deadline. In a study of more than 1,350 participants, Cepeda and colleagues varied the delay between initial learning and a review, then the delay until the final test. The spacing associated with the best performance increased as the final deadline became more distant.

This study concerned learning facts in an experimental protocol. It does not therefore provide a ready-made schedule for every collège (lower secondary school) or lycée (upper secondary school) student.

Situation	Useful logic	Common mistake
Test in a few days	Several short reviews spread out before the deadline	Cramming everything the night before
Chapter to retain throughout the term	Come back to it over several weeks	Stopping all revision after the test
Cumulative vocabulary	Space out more the items that are regularly recalled	Reviewing all cards at the same frequency
Poorly understood concept	Go back to the explanation or a guided example first	Mechanically scheduling a misunderstanding that has not been corrected

Adapting the next recall to what has just happened

A practical rule is to distinguish three situations:

1. No answer or a wrong answer: correct, then bring the concept back relatively soon.
2. A correct but hesitant answer: keep a moderate delay.
3. A precise answer that is regularly recalled: space out the next review more.

This is not a scientific algorithm. It is simply a way of avoiding spending as much time on knowledge that is already solid as on points that are still shaky.

4. Combining spaced repetition and active recall

Spacing revision answers the question when should I revise?. Active recall answers the question what should I do while revising?.

A revision session can follow five steps:

1. hide the material;
2. try to retrieve the information;
3. check the answer;
4. correct the mistake precisely;
5. schedule a new attempt after a delay.

The experiment by Karpicke and Blunt published in Science also showed, among university students learning scientific texts, an advantage of retrieval practice over several forms of restudying in later tests. It did not concern a school flashcard schedule, but it shows why the activity carried out during each review matters as much as its date.

Our article on active recall sets out this logic in detail. The one on using flashcards well shows how to turn a lesson into short, reusable questions.

5. An example four-week schedule

The following schedule is a teaching example, not a scientific prescription. Imagine that a chapter is learnt on Monday and the student wants to retain it for several weeks.

When	Work	What is checked
Day 0	Understand the lesson, then do a first recall	Initial understanding
Day 2	Questions, flashcards or recalling without notes	The first things forgotten
Day 6	A new recall, then an application	Availability after several days
Day 13	A quiz mixing this concept with others	The ability to distinguish and select it
Day 27	Overall recall, then targeted review	What is still available after several weeks

If a concept is forgotten on day 6, there is no need to wait mechanically until day 13. It can be corrected and then retested sooner. Conversely, a definition recalled easily several times can wait longer.

The schedule must remain manageable

A method can become counterproductive if organising it takes longer than the learning itself. Creating hundreds of cards, piling up overdue reviews and then trying to catch up on everything easily leads to giving up on the system.

So start with the important knowledge. A few targeted recalls that are actually carried out are better than an extremely precise programme that lasts only four days.

6. Adapting spaced repetition to the subject

Languages: producing vocabulary, not just recognising it

Words and expressions lend themselves naturally to spaced reviews. But recognising a foreign word when it appears is not the same as being able to produce it. It is therefore useful to revise in both directions and to use some words in sentences.

History, SVT (Life and Earth Sciences) or SES (Economic and Social Sciences): reviewing facts and links

Dates, definitions and vocabulary can be recalled directly. For mechanisms, add questions such as “why?”, “what are the causes?”, “what is the difference between these two concepts?” or “reconstruct the stages of this phenomenon”.

French: memorising, then reusing

Spacing can be used for literary reference points, analysis vocabulary, certain quotations or elements of method. It does not replace writing a paragraph, building an essay plan or analysing a new text.

Maths: spacing problems, not just formulas

Spaced repetition can also work in maths, provided that mathematical tasks are spaced. Periodically recalling a formula is useful, but it does not prove that the student can recognise the situation in which it applies.

A meta-analysis published in 2025 on spacing and retrieval practice in mathematics found a robust, small-to-medium effect of spaced practice compared with massed practice across 27 studies and 53 effect sizes. However, the available data on the effect of retrieval compared with restudying in mathematics were less conclusive.

For maths, it therefore makes sense to redo a type of exercise a few days later, then mix it with other problems.

7. The limits of automatic apps

A spaced repetition app can solve a real organisational problem: keeping a history of reviews, showing certain cards at the right time and avoiding having to manage several piles by hand.

But it does not automatically know whether:

• a card is badly worded;
• the answer was understood or merely recognised;
• a memorised concept can be applied;
• a difficulty comes from a missing prerequisite;
• the student can write, reason or solve a complete problem.

The sophistication of the schedule should therefore not be confused with the quality of learning. Research strongly supports the general principle of spacing, but it does not justify presenting any particular algorithm as optimal for every student, every subject and every deadline.

Do not turn the algorithm into the goal

If an app indicates that a card should be reviewed when the student no longer even understands the concept, you need to step out of the simple memorisation cycle and go back to the explanation.

Using Learn pack flashcards in a spaced way

On NOEO, the student can create a Learn pack on their own initiative. It contains a revision sheet, flashcards and a quiz. It can be created from a completed session, from several completed sessions when they involve the same tutor and the same subject, or freely from content added by the student.

The flashcards can naturally be revisited at several moments. A sensible way to use them is to attempt the answer before revealing it, identify the difficult cards and then come back to them in later sessions. The quiz then makes it possible to mix several pieces of knowledge and bring other weaknesses to light.

There is no need to claim that a particular algorithm automatically organises optimal repetition. The educational value comes first and foremost from the succession of attempts and the way the student deals with their mistakes.

The Learn pack is not generated automatically after each session. Nor does it contain the exercises sent by the tutor via the Session report. When the quiz reveals a difficulty, our article on revision quizzes and analysing wrong answers helps distinguish between forgetting, confusion and an application problem.

A simple protocol to get started

1. Choose the knowledge that genuinely needs to remain available.
2. After learning it, try to retrieve it without looking.
3. Correct misunderstandings immediately.
4. Come back to this knowledge a few days later.
5. Bring back sooner what has been forgotten and space out what remains solid.
6. For subjects that require it, add exercises, written work or new applications.