Short answer first: review a new flashcard after one day, then after three days, a week, two weeks and a month. Get it right each time and the card is safe for two to three months. Get it wrong and it goes back to day one. Add ten to twenty new cards a day, and expect the reviews to take about fifteen minutes a day once the deck has settled.
That is the whole schedule. The rest of this article is what each step means in practice, what to do when a card fails, how many cards you can add without drowning, and the research the numbers come from. If you want the background on why spacing works at all, start with spaced repetition and come back here for the timetable.
What is the schedule?
The schedule is five reviews at growing gaps, each one a real attempt to answer before you look.
| Review | When | If you get it right | If you get it wrong |
|---|---|---|---|
| 1 | Day 1 (the day after you made the card) | Next review on day 3 | Retry in the same session, review again tomorrow |
| 2 | Day 3 | Next review on day 7 | Back to day 1 |
| 3 | Day 7 | Next review on day 14 | Back to day 1 |
| 4 | Day 14 | Next review on day 30 | Back to day 1 |
| 5 | Day 30 | Next review in two to three months | Back to day 1 |
"Day 1" is counted from the day you made the card or first learned the material, not from the lecture. The first review is the one most people miss, and it is the one that matters most, because the forgetting curve is steepest in the first day. In Murre and Dros's 2015 replication of Ebbinghaus, most of what was going to be lost was lost within the first day, and the curve flattened after that.
Why these gaps and not others
The gaps roughly double each time, which is the pattern that falls out of the research: every successful recall makes the memory more durable, so the next gap can be longer. The exact numbers are not sacred. Cepeda and colleagues found in a 2008 study with over 1,350 participants that at any given test delay, performance rose as the gap grew and then fell again, with a broad plateau near the top rather than a sharp peak. A review a day early or two days late costs little. Skipping one costs a lot.
What do you do when you fail a card?
Send it back to the start. A miss means the gap was too long for that card, so the fix is a shorter gap, not a slightly shorter one.
Retry in the same session. Read the answer, put the card aside, and answer it again five minutes later. Do not count the second attempt as a hit; it is the first step of relearning.
Schedule it for tomorrow. The card is treated as new. It goes through day 1, day 3 and the rest of the ladder again. This feels like a punishment and is not one; a card that fails at day 14 and gets nudged to day 10 will usually fail again.
Rewrite cards that fail three times. A card you keep missing is normally a bad card, not a bad memory. It asks two things, or it asks something you never understood in the first place. Split it, or go back to the notes it came from and fix the understanding before you fix the card. The Anki manual describes the same mechanics: pressing Again on a review card is a lapse, and the card goes through relearning steps before it becomes a review card again.
How many new cards a day?
Ten to twenty. The limit is set by the reviews the new cards generate, not by how many you can write. Every card you add today comes back on day 1, 3, 7, 14 and 30, so a deck that grows by twenty a day settles at a review load of around two hundred a day. That number is from the Anki manual, which warns about exactly this, and it matches what most students find: twenty new cards a day is sustainable for one module, and too many for three.
A practical rule for a full course load: ten new cards a day per module you are actively studying, and no new cards on days when the review pile is over a hundred. Flashcards from your notes covers how to keep the number down by writing one card per idea rather than one per fact.
How long should a session be?
Fifteen to twenty minutes a day once the deck is mature, plus the time to make new cards. A review takes a few seconds when you know the answer and half a minute when you do not, so two hundred reviews is about fifteen minutes. Do it at the same time every day; the schedule only works if the day 3 review happens on day 3. Put it in your study schedule as a named slot, not as a general intention.
What does the research say?
Three findings underpin the schedule.
Ebbinghaus: forgetting is fast at first, then slow
In the 1880s Hermann Ebbinghaus memorised lists of nonsense syllables and measured how much he had to relearn after gaps from twenty minutes to a month. He found that most forgetting happened within hours, and that the rate slowed sharply after the first day. Murre and Dros repeated the experiment in 2015 with the same method and intervals and got the same curve. The first review has to come early because that is when the loss is largest.
Cepeda and colleagues: the best gap depends on the test date
The strongest evidence for spacing is a 2006 review by Cepeda, Pashler, Vul, Wixted and Rohrer, which pulled together 839 assessments of distributed practice from 317 experiments. Spaced study beat massed study almost everywhere, and the gap that produced the best retention grew as the time until the test grew. Their 2008 follow up put numbers on it with over 1,350 participants: for a test about a week away, a gap of about a day was best; for a test a year away, a gap of about three weeks. That is why the schedule above stretches out as it goes, and why an exam in June wants longer gaps than a quiz on Friday.
Kornell: spacing beats cramming with real flashcards, and students cannot feel it
Kornell's 2009 study is the one that speaks most directly to flashcard users. Participants learned vocabulary either as one large stack of cards, so each card came round again after many others (spacing), or as four small stacks studied one after another (cramming). The large stack won for 90 per cent of participants. After the first session, 72 per cent of them believed the small stacks had worked better. Cramming feels more effective because the card is still fresh when it comes back. That feeling is exactly what the schedule is there to override.
The mechanism on each review is active recall: the memory is strengthened by the retrieval, not by the exposure. Reading the answer before you try to produce it turns the review into rereading, and the schedule stops working.
How do you fit the schedule around an exam?
Count back from the exam date and choose the ladder that fits.
| Time until the exam | Reviews per card | Gaps |
|---|---|---|
| Six weeks or more | 5, then a final pass | Day 1, 3, 7, 14, 30, then the week before |
| Three to four weeks | 5 | Day 1, 3, 7, 14, 21 |
| Two weeks | 4 or 5 | Day 1, 2, 4, 7, 12 |
| One week | 3 | Day 1, 3, 6 |
| Two days | 1 | Tomorrow, which is cramming |
The shorter the ladder, the faster the material fades after the exam, which matters if the module builds on it. How to fit the ladder into the final three weeks is covered in exam preparation.
Paper boxes or an app?
Both work. On paper, use five boxes labelled 1, 3, 7, 14 and 30. A hit moves the card one box forward, a miss moves it back to box one, and you review each box on its day. It works well for one module and becomes a chore for three. An app does the arithmetic, shows you today's pile, and keeps the misses honest, because it will not let you quietly skip a card you dislike. Anki is the standard choice. Notibo builds the cards from your lecture recordings and uploaded files and schedules them on the same kind of ladder, which removes the step of writing the cards but not the step of answering them every day.