The forgetting curve: what the data show and how to flatten it

What the forgetting curve is

The forgetting curve describes how memory for newly learned material declines over time: steeply at first, then more and more slowly. It comes from Hermann Ebbinghaus, who in 1880 and 1885 tested himself on lists of nonsense syllables and measured how much he retained after intervals from 20 minutes to 31 days.

It is one of the most cited findings in psychology and one of the most misquoted. Popular posts claim you "forget 70% of what you learn within 24 hours" or similar. Those numbers do not describe what Ebbinghaus measured, and they do not describe how you forget the content of a certification course. This page sets out what the curve actually shows, using a modern replication, and what you can do about it.

Why it matters for exam candidates: most certification preparation runs for six weeks or longer. The topics you study first are the ones most at risk by exam day, unless you come back to them. The spacing research tells you when those returns pay off most.

The modern replication: Murre and Dros (2015)

Because Ebbinghaus tested only himself, Murre & Dros, 2015 repeated his experiment as closely as they could. One subject spent 70 hours learning lists of nonsense syllables and relearning them after 20 minutes, 1 hour, 9 hours, 1 day, 2 days, 6 days or 31 days. Their conclusion: the results were similar to Ebbinghaus' original data, and the classic forgetting curve has indeed been replicated.

What "savings" means

Ebbinghaus did not measure what percentage of syllables he could recall. He measured savings: how much less time it took to relearn a list to the point of one perfect recitation, compared with learning it the first time. A savings score of 0.33 means relearning took about a third less time than the original learning. Murre and Dros used the same method.

The table on this page shows the savings scores reported in the paper for Ebbinghaus and for the replication. For Ebbinghaus, savings fell from 0.582 at 20 minutes to 0.337 at 1 day and 0.211 at 31 days. The replication followed a similar path, with a lower value at 31 days.

Why "you forget 70% in a day" is wrong

• Savings is not recall. A savings of 0.337 at one day means relearning took about two thirds of the original time. It does not mean two thirds of the knowledge was gone.
• The material was meaningless on purpose. Ebbinghaus chose nonsense syllables because they have no meaning to attach them to. The curve is a poor guide to how fast you forget a concept you understand and have practised.
• It is one person at a time. Both the original and the replication are single-subject studies. They show the shape of forgetting, not a population average.

The curve is not smooth: the 24-hour bump

The most interesting finding of the replication is a detail. Murre and Dros concluded the forgetting curve is not completely smooth: it most probably shows a jump upward starting at the 24-hour point. Their analysis found relatively less forgetting at 1 day, 2 days or both than a smooth curve would predict, in Ebbinghaus' data and in the replications. In the replication, savings at 1 day (0.317) was even higher than at 9 hours (0.276).

The authors note that current research on sleep and memory would predict such a jump after a night of sleep, but they are careful: for this type of experiment, the role of sleep remains to be established.

Independent evidence on sleep is stronger. A meta-analysis by Newbury et al., 2021 found total sleep deprivation impairs memory, with a larger effect before learning (g = 0.621) than after learning (g = 0.277), though the authors note publication bias and low power. A later meta-analysis found that restricting sleep to 3 to 6.5 hours, compared with 7 to 11 hours, impairs memory formation (g = 0.29, a small effect), and that missing some sleep has similar consequences for memory to not sleeping at all (Crowley et al., 2024).

Even quiet waking rest seems to help: brief periods of rest after learning facilitate consolidation of new memories (Wamsley, 2019). Ten quiet minutes after a study block may do more than immediately scrolling through your phone.

How to flatten your forgetting curve

You cannot stop forgetting, but you can change its slope. Two techniques have the strongest evidence.

1. Retrieve instead of rereading

Retrieval practice often produces large gains in long-term retention compared with repeated studying (Roediger & Butler, 2011). Each successful retrieval makes the memory more durable. See active recall for practical methods.

2. Space your reviews

A meta-analysis of 317 experiments found that the best gap between study sessions grows with the time until the test (Cepeda et al., 2006). Each review, at the right moment, resets the curve and makes the next decline slower. That is the logic of spaced repetition.

The replication even contains a hint of this. Murre and Dros found an unusually low savings score at 31 days, and one explanation they offer is that the lists for that interval were all learned early in a short, massed period, while learning for other intervals was more widely spaced. They present this as speculation, but it fits the spacing research.

3. Protect sleep, especially before and after learning

Based on the sleep meta-analyses above, all-nighters before an exam work against you twice: they impair memory for anything learned that night and for what you learned in the days before.

The good news hidden in "savings"

The savings method carries an encouraging message that the popular version of the forgetting curve leaves out. Even at 31 days, when Ebbinghaus could not have recited a list, relearning it was still faster than learning it from scratch. Memory that you cannot recall is often not gone: it makes relearning quicker.

For exam candidates, that means the hours you spent on a domain in week one are not wasted if you have "forgotten" it by week five. A short review brings it back faster than the first time. The mistake is to leave that review until the night before the exam, when there is no time to retrieve it again on a later day.

It also means you should not panic when a practice test shows you have forgotten early material. That is normal, expected and fixable. Plan reviews instead of hoping memory will hold.

The practical fix is a calendar, not willpower. Put review sessions into your study schedule the moment you finish a topic, so the reviews happen whether or not the topic still feels fresh. For high-volume facts such as ports, acronyms and definitions, a flashcard box system such as the Leitner system does the scheduling for you: cards you miss come back soon, cards you know come back later. Reviews will feel effortful when some forgetting has happened. That effort is a desirable difficulty, not a sign you are failing.

How to use this for a certification exam

A protocol that fights the forgetting curve over a typical 6 to 8 week preparation period:

1. Same day. After each study block, close the material and write down everything you remember for 5 to 10 minutes, then check. Rest quietly for a few minutes before the next block.
2. Next day. Start the session with 10 questions on yesterday's topic, from memory.
3. About one week later. 10 to 15 questions on the topic again, mixed with other topics.
4. Two to three weeks later. A mixed set of 30 questions across everything learned so far. Retake our free CompTIA A+ practice test or Security+ practice test every two weeks to see which domains are slipping.
5. Final week. Daily mixed sets of 20 to 30 questions, with extra time on the domains your last practice test flagged.
6. Sleep. Aim for a full night's sleep throughout, and especially in the last three nights before the exam. No all-nighters.

For the timing logic behind these intervals, see spaced repetition. For the complete exam plan, see how to study for exams.