What Is the Forgetting Curve? Leave What You Learned Unreviewed and It Fades Fast at First, Then Slowly

The forgetting curve describes something plain: leave something unpractised after studying it and the part you can still recall drops quickly at first, then more and more slowly. It is not a prediction about you personally — it is a trend averaged across a large batch of items. That is also why "learn it and walk away" is close to wasted effort: the steepest losses happen in the first few days.

Two things up front. The curve describes a population of items, not a verdict on the one word in front of you, and no figure for how many words you will have forgotten tomorrow is computed anywhere in the app. We are not going to invent one either.

1. The definition, in one sentence

A forgetting curve plots how much is still retrievable against how long ago you last studied it: retention falls fastest over the first hours and days, then decelerates instead of sliding at a constant rate.

The forgetting curve is a description of a population. It is not a timetable for a person.

That sentence is doing most of the work here. A curve is a summary of measurements taken under particular conditions: particular material, particular people, particular delays, and one particular way of deciding whether something counts as remembered. Change any one of those and you get a different curve with the same name. The popular version treats it as a law, as though every item you learn sits on one fixed trajectory and all you have to do is intercept it. Nothing in the original work supports that reading.

2. How it was measured in the first place

It comes from Hermann Ebbinghaus, who published it in 1885. Three details about how he got it are worth more than the curve itself:

• The only subject was himself. Over several months, one person took part: Ebbinghaus.
• The material was deliberately meaningless. He invented nonsense syllables — consonant-vowel-consonant strings like zof and wux — built specifically to carry no existing associations.
• What he measured was relearning, not recall. He used the saving method: relearn a list to criterion and count how many fewer runs it took the second time. The vertical axis is effort saved, not exam questions answered correctly.

We do not repeat the percentages that circulate in articles about this curve. They cannot be traced to a source that holds for vocabulary, and the figures quoted from one article to the next do not agree with each other. The shape — fast at first, then slower — has held up; the specific numbers should not be carried over to the words you are studying.

3. What it means when you apply it to vocabulary

• The first few days matter most. The steepest drop comes right after learning, so a new word is better met again within the first few days than left for a week.
• Meaningful material behaves differently. Ebbinghaus used syllables with nothing to hang an association on, while the words you study usually come with collocations, example sentences and context — close to the opposite of his material.
• "Forgotten" is not a single state. On the same word you might be able to write it, only recognise it, or manage neither. Without naming which test you mean, there is no single curve — which is why translation, dictation and speaking are three separate skills here.
• Review pushes it back up. Every successful retrieval makes the next drop come later, which is exactly what spaced repetition does.

4. Three things you can do now

• Bring new words back several times in the first few days. The steepest losses are right after learning, so do not push the first reunion a week out. Forgetting a word the next day is normal — see Forgetting It the Next Day.
• Keep in touch during dead time. Put word music on while commuting, queueing or doing housework, so the words keep passing your ear. It costs far less than a catch-up session at the weekend — see Word Music.
• Keep the hard ones in Flags. Flag words you keep failing to recall and they are gathered in one place, ready for a focused pass later — see Turning Mistakes Into a Signal.

5. What VocaX does here

It sets the rhythm on which each word comes back: words you have studied are brought back as they start to slip, so you never have to track which word is due on which day. Your job is to open the app each day and finish the batch it has lined up. Which words are due today has already been decided for you.

The steepest drop is in the first few days. Rather than betting on your own memory, let each word come back by itself just as it starts to slip.

6. Where it stops helping

• It is a population trend, not a prediction about you. The curve says how a batch of items fares on average, not that you will have forgotten eight words in three days.
• There is no single correct "best moment to review." Change the test and you change the curve; no amount of curve yields one hour that everyone should obey. Treat any packaged "golden revision point" with suspicion.
• It will not tell you how many words you will forget tomorrow. No such figure exists in the app, and we cannot produce one.
• It does not choose your words. The curve says that what you studied will fade; it says nothing about what to study next — see Choosing a Word List.
• It is an iPhone app. VocaX is iOS only; there is no Android version.

Treat the forgetting curve as one reminder and it serves you well: leave something unpractised and it fades. Treat it as a timetable to execute precisely and it does not.

FAQ

Was the forgetting curve measured on real words?

No. Ebbinghaus published it in 1885, the only subject was himself, and the material was nonsense syllables he invented — consonant-vowel-consonant strings like zof and wux, built to carry no meaning. What he measured was relearning: study the list again and count how many fewer runs it took the second time.

Are the percentages quoted online trustworthy?

We do not repeat them, because they cannot be traced to a source that holds for vocabulary and the figures quoted from article to article disagree. The shape — fast at first, then slower — has held up; those specific percentages should not be carried over to the words you are studying.

Is there an optimal moment to review?

There is no single answer. Change the test and you change the curve: recall, recognition and relearning draw different lines over the same delay. The reliable approach is to let each word come back as you are about to lose it, rather than chasing a fixed number of hours — see What Is Spaced Repetition.

I have not opened the app for days. Is it all lost?

No. Come back after a gap and some of the progress you had banked will have slipped — that is normal. Just work through the review that piled up; there is nothing to reset. VocaX will not nag you about it either — see Why We Do Not Nag You.

What time of day is best for reviewing?

There is no universally best hour; the one you can actually keep is the best one. What matters more than the slot is finishing each day's review rather than letting it pile up until the weekend.

Axiom One LLC — VocaX. Figures current as of 15 September 2026.