Do you really forget 70% of what you learn in 24 hours?

For disconnected material learned once, losses in the first day are steep and figures in that range come from Ebbinghaus-style experiments with nonsense syllables. Meaningful, well-connected material decays more slowly. The reliable claim is that the loss is front-loaded, not that a precise percentage applies to everything.

What is the fastest way to beat the forgetting curve?

One short retrieval attempt the day after you learn something, then a few progressively spaced attempts over the following weeks. Retrieval matters more than duration: two minutes of recall beats ten minutes of rereading.

Is the forgetting curve the same for everyone?

The shape is general; the slope varies with sleep, prior knowledge, how meaningful the material is and how it was encoded. Someone who already knows a subject forgets new details in it far more slowly, because they have more places to attach them.

Does sleep affect the forgetting curve?

Yes. Consolidation happens substantially during sleep, and studies consistently find poorer retention after short or disrupted sleep. A late-night cram followed by four hours of sleep is close to the worst possible combination.

Learning Science

The Forgetting Curve: Why You Lose 70% of What You Learn in 24 Hours

Marisol Vega

Psychology & Habits Writer

Last updated: August 2026

9 min read

TL;DR

Memory for freshly learned material drops steeply within the first day and then flattens. The curve is not a life sentence: reviewing at increasing intervals, testing yourself instead of rereading, and connecting new facts to things you already know all reduce the slope. The single highest-value habit is a short second pass on day two.

You have had the experience. A talk that felt clarifying on Tuesday is, by Wednesday afternoon, one remembered slide and a vague sense of enthusiasm. That is not a personal failing. It is the most reliably documented shape in the psychology of memory.

What the forgetting curve is

In the 1880s the German psychologist Hermann Ebbinghaus tested memory on himself using lists of nonsense syllables, learning them to perfect recall and then measuring how much effort it took to relearn them after various delays. Plotted, the results form a steep drop shortly after learning that gradually levels off. Most of what is lost is lost early.

The often-quoted figure is that you lose most of new material within twenty-four hours. Treat the exact percentage carefully. Ebbinghaus used meaningless syllables, one subject, and no attempt to connect the material to anything. Meaningful material learned with context decays more slowly. The shape is robust; the numbers are illustrative.

Why forgetting is not a bug

A memory system that kept every detail with equal strength would be useless. You do not want yesterday's parking space competing with your own address. Forgetting prunes what has not proven useful, and 'useful' is inferred from repetition and retrieval. When you go back to something, you signal that it matters. When you do not, the system correctly assumes it did not.

This reframes the whole problem. You are not fighting a defect. You are sending the wrong signal by only ever encountering material once.

The four things that flatten the curve

1. A second pass on day two

Because the drop is steepest early, the earliest review is worth the most. A two-minute revisit the day after learning something has a disproportionate effect on what survives a month. If you only adopt one habit from this article, adopt this one.

2. Spacing, not massing

Three ten-minute sessions across three days beat one thirty-minute session, even though the total time is identical. Each retrieval after a gap forces reconstruction rather than recognition, and reconstruction is what strengthens the trace. The mechanism and how apps implement it is covered in spaced repetition, explained like a museum tour.

3. Retrieval instead of rereading

Rereading feels productive because familiarity rises. Familiarity is not recall. Cover the page and say the answer out loud, or write the summary from memory before checking. This is the testing effect and it is one of the largest, most replicated findings in learning research.

4. Connection

Isolated facts decay fastest. A fact attached to a story, a cause or a face has multiple retrieval routes. This is why narrative learning outperforms lists: if you know why Constantinople's walls failed in 1453, the date arrives free, which is how The Day the Middle Ages Ended is structured.

A practical review schedule that fits a real week

When	What to do	Time
Immediately after	Say the one-sentence version out loud	30 seconds
Next day	Recall it before checking your notes	2 minutes
Three days later	Explain it to someone or write it from memory	3 minutes
One week later	Quick self-test on the parts you got wrong	3 minutes
One month later	One retrieval attempt, no rereading first	2 minutes

Total time under fifteen minutes per item, spread across a fortnight.

Common mistakes that make forgetting worse

Highlighting as a study method. Highlighting marks text for future-you and does no encoding work now.
Reviewing while the material is still fresh. If recall is effortless, the review is not doing much. Some difficulty is the point.
Learning in one long block before a deadline. Cramming produces a real short-term peak and an unusually steep drop afterwards.
Consuming more instead of returning. Another new summary feels like progress and adds another item to the pile of things you will not remember.
Treating recognition as knowledge. 'I have seen this' and 'I can explain this' are different states.

What this means for daily learning apps

Any app that shows you material once and never asks about it again is working against the curve, however good the content is. The features that matter are unglamorous: does it bring earlier material back, and does it ever make you answer something? That is the lens we use when we compare products, including in our ranking of micro-learning apps, where we disclose that MindSnap is our own app.