Ebbinghaus Forgetting Curve Explained (and How to Beat It)

The forgetting curve shows how quickly you lose newly learned information when you don't review it: steeply in the first hours, then more slowly over days and weeks. Hermann Ebbinghaus measured it on himself in 1885, and a 2015 replication found nearly the same shape.

The curve is real, but most versions online get the numbers wrong. Below you'll find what Ebbinghaus actually measured, his data next to the replication, why "you forget 70% in 24 hours" is a misreading, the formulas, and a review schedule that keeps the curve flat.

What is the forgetting curve?

The forgetting curve is a graph of how much of something you've learned is still available to you as time passes without review. It starts high, drops steeply over the first hours, and then flattens into a long, slow decline.

Two features matter for studying. First, most forgetting happens early, so a quick review in the first day does more than a long one a week later. Second, the curve describes material you never revisit. Each time you successfully recall something, the next curve is shallower, which is the idea behind spaced repetition.

How did Ebbinghaus measure forgetting?

Ebbinghaus was his own and only subject. He built lists of nonsense syllables, made-up consonant-vowel-consonant combinations with no meaning, so that prior knowledge couldn't help him. For the forgetting curve experiment he used series of 13 syllables, repeated them until he could recite them without error, and timed how long that took.

Then he waited a set time, from 20 minutes to 31 days, and relearned the same lists. His measure was savings: how much less time relearning took than the first learning.

For example, if a list takes 20 minutes to learn and 13 minutes to relearn the next day, the saving is 7 minutes, or 35%. A saving of 100% would mean no relearning was needed; 0% would mean starting from scratch.

The forgetting curve graph: 1885 vs 2015

In 2015, Jaap Murre and Joeri Dros repeated the experiment as closely as they could. One subject, Dros himself, spent about 70 hours learning lists of nonsense syllables and relearning them after the same delays. The two curves sit close together:

What the two curves agree on:

• The steepest drop comes first. Most of the loss happens within the first hour or so.
• After a day, forgetting slows. Ebbinghaus still saved 21% of his learning time a full month later.
• There's a bump at 24 hours. In the replication, savings after one day were higher than after nine hours, and in Ebbinghaus's data they barely fell. Earlier researchers put this down to sleep. Murre and Dros say it fits sleep research, but their data can't prove sleep caused it.

The replication's 31-day point is much lower (4%). The authors note that those lists were all learned early in the experiment, partly in a more massed way, and that learning times crept up over the study. Correcting for that drift raised the value to about 14%, still below Ebbinghaus's 21%.

How much do you forget in 24 hours?

You'll often read that people forget 50% of new information within an hour and 70% within a day. Those numbers come from misreading Ebbinghaus's savings scores.

His 33.7% saving after one day means relearning took about two-thirds as long as learning the first time. It does not mean two-thirds of the list was gone. Savings measures relearning effort, not how much you can recall. When Murre and Dros also scored how many syllables were recalled correctly, that curve was much shallower than the savings curve.

The material matters too. Nonsense syllables were chosen to be meaningless. Ebbinghaus himself found that a stanza of Byron's Don Juan, about 80 syllables long, took him roughly a tenth of the repetitions he estimated an equally long nonsense list would need.

Meaningful, connected course material behaves differently from random syllables, so no single percentage tells you how much of tonight's lecture you'll lose by tomorrow. What holds up is the shape: fast loss early, slower loss later. Our guide to why we forget covers the causes.

The forgetting curve formula

There isn't one true formula. Ebbinghaus fitted this equation to his 1885 data:

b = 100k / ((log t)^c + k), where b is the percentage saved, t is minutes since learning, log is base 10, k = 1.84, and c = 1.25.

He was careful about it, calling the formula "no other value than that of a shorthand statement" of his own results.

Two other forms are common today:

• Exponential: R = e^(−t/S), where R is the probability of recall, t is time, and S is the memory's strength. It's the simplest model and appears in many explainers.
• Power law: the rate of forgetting keeps slowing as time passes. Murre and Dros compared several equations on four data sets and found a power function with a 24-hour bump and a two-stage "memory chain" model fit about equally well.

Modern flashcard schedulers often use a power curve. FSRS-4.5 and FSRS-5, for example, predict recall as R = (1 + 19/81 × t/S)^−0.5, where S is "stability": the number of days until your chance of recalling the card drops to 90%. By that formula, a card with a stability of 10 days has a 77% chance of recall after 30 days and about 55% after 100 days. See FSRS vs SM-2 for how apps use this to schedule reviews.

How do you beat the forgetting curve?

1. Review within a day

The curve is steepest at the start, so the first review gives you the most for the least effort. Ten minutes of recall the evening after a lecture, or the next morning, beats an hour the following week.

2. Recall instead of rereading

In a 2006 study, students who read a passage once and then took a recall test remembered 56% of it a week later, compared with 42% for students who read it twice. Rereading felt better at the time and lost later. Use blank-page recall, flashcards, or practice questions, as described in what is active recall.

3. Space the reviews out

Each successful recall makes the memory more durable, so the gap before the next review can be longer. Across 271 comparisons in a 2006 review, spaced study beat the same amount of massed study on average at every test delay. See the spacing effect for how far apart to space sessions for your exam date.

4. Make it mean something

Connect new facts to things you already know, explain why they're true, and use examples. Meaningful material is learned far faster than isolated facts, as Ebbinghaus's poetry test showed.

5. Sleep on it

Sleep helps consolidate what you learned during the day. Cutting sleep to study more works against you; see sleep and memory consolidation.

A review schedule that flattens the forgetting curve

Use this for each lecture, chapter, or set of cards. Every review is a recall attempt first, then a check.

If you blank at any review, go back to a short gap and build up again. A flashcard app does this per card automatically; on paper, a Leitner box does the same job.