How much do you forget after 20 minutes, 1 hour, 24 hours, 1 week, and 1 month — and what you can do about it
| Time since learning | % forgotten | % retained |
|---|---|---|
| 20 minutes | 42% | 58% |
| 1 hour | 56% | 44% |
| 9 hours | 64% | 36% |
| 24 hours | 67% | 33% |
| 2 days | 72% | 28% |
| 6 days (1 week) | 75% | 25% |
| 31 days (1 month) | 79% | 21% |
Source: Ebbinghaus, H. (1885). Über das Gedächtnis. Based on savings scores from nonsense syllable experiments.
The forgetting curve is a mathematical model of how quickly memory fades over time without reinforcement. German psychologist Hermann Ebbinghaus discovered it in 1885 through a series of meticulous self-experiments using nonsense syllables (meaningless letter combinations like "WID", "ZOL", "PAG").
Rather than measuring what he could freely recall, Ebbinghaus used a "savings score": he measured how much less time it took to relearn the same list after a delay compared to the original learning session. A 33% savings score at 24 hours means relearning took 33% less effort than the first time — implying roughly 33% of the original memory trace was still accessible.
This is why the "exact percentages" you see cited vary slightly across sources. The 58/44/33/25/21% figures in the table above are the most accurate representation of Ebbinghaus's original data. The commonly cited "70% forgotten in 24 hours" is a rounded approximation.
The most important insight from the curve is its exponential shape: forgetting happens very fast at first, then slows dramatically.
Notice that the drop from 24 hours (67% forgotten) to 1 month (79% forgotten) is much smaller than the drop in the first hour. This means that if you can survive the first 24–48 hours with a memory intact, it becomes much more durable. Spaced repetition exploits this directly.
Ebbinghaus's original curve was measured under controlled conditions using meaningless syllables. Real-world forgetting rates vary significantly based on:
Information with emotional significance or connections to existing knowledge is retained far longer. You'll forget a random phone number much faster than your childhood address.
Passive re-reading produces shallow encoding. Active recall during initial learning — generating answers, making connections — creates stronger memory traces that decay more slowly.
Sleep plays a critical role in memory consolidation. Studies show that memories reviewed before sleep are retained at roughly 2× the rate of memories reviewed at other times.
High stress (like cramming under pressure) can accelerate forgetting by impairing consolidation. Moderate arousal aids memory; excessive stress harms it.
Experts forget domain-specific information much more slowly because new material can attach to rich existing knowledge networks. This is why building foundational knowledge early pays dividends.
Ebbinghaus didn't just discover the problem — he also found the solution. He showed that reviewing material at precisely spaced intervals dramatically reduced the total effort required to maintain memory. Each review session "resets" the curve, and subsequent curves decay more slowly.
Research-backed review intervals for moving information into long-term memory:
| Review session | When to review | Effect on the curve |
|---|---|---|
| 1st review | Same day (within 24 hours) | Prevents the steepest drop; retention ~80% |
| 2nd review | 3 days later | New curve decays from higher baseline |
| 3rd review | 7 days later | Memory begins entering long-term storage |
| 4th review | 14 days later | Interval can now be extended safely |
| 5th+ reviews | 30+ days (growing intervals) | Near-permanent retention with minimal effort |
The key insight: you don't need to review more — you need to review at the right time. Reviewing too early wastes effort. Reviewing too late means you've forgotten and must relearn. The sweet spot is reviewing just as you're about to forget — which is exactly what spaced repetition algorithms calculate for you.
A student who reviews material with proper spaced repetition can retain 90%+ of what they learned using only a fraction of the total study time compared to cramming.
StudyCards AI uses spaced repetition science to automatically schedule reviews at the exact intervals shown above — adapted to your personal forgetting rate. Upload your notes and it generates the flashcards for you.
According to Ebbinghaus's savings scores, approximately 67% of newly learned material is forgotten within 24 hours without any review. Only about 33% remains accessible. This is why reviewing material the same day you learn it is so important — it prevents the steepest part of the forgetting curve.
After approximately 6 days (one week), about 75% of material is forgotten, leaving only 25% retained. The curve is flatter at this point — the rate of forgetting has slowed significantly compared to the first 24 hours.
The "70% forgotten in 24 hours" is a commonly cited approximation. Ebbinghaus's actual data showed ~67% forgotten (33% retained) at 24 hours. The number has been rounded and misquoted across decades of popular science writing. The 67% figure is more accurate for the specific conditions of Ebbinghaus's experiments (meaningless syllables, single learning session).
Ebbinghaus's exact percentages came from memorizing nonsense syllables — one of the worst-case scenarios for memory. Real educational material (which has meaning, structure, and connects to what you already know) tends to be forgotten more slowly. However, the shape of the curve — fast initial drop, then leveling off — applies broadly to most declarative learning.
Research generally suggests that 4–6 well-spaced review sessions can move most material into very long-term retention. Using intervals of roughly 1 day, 3 days, 7 days, 14 days, and 30 days is a practical starting point. After that, review intervals can extend to months or years for most material.
References
Generate Anki flashcards free