Most students already know the advice: review a little and often instead of cramming. The hard part is turning it into a plan. Which cards do you look at today? How long until you see a card again? What happens when you get one wrong on a Thursday and miss the weekend? Spaced repetition answers those questions with a rule simple enough to follow on paper, and this article walks through one exact version of it, the five-box Leitner schedule used by the flashcards tool, with real dates you can check.
The idea in one paragraph
Memory fades after you learn something, and a review interrupts that fading. Each successful review lets you wait longer before the next one. So instead of seeing every card every day, you see a card again soon after you first learn it, then after a slightly longer gap, then a longer one still. Cards you find easy drift out to long intervals and stop taking your time; cards you find hard keep coming back until they stick. The forgetting side of this goes back to Hermann Ebbinghaus's experiments on himself, published in 1885 as _Über das Gedächtnis_. The card-box method comes from Sebastian Leitner's 1972 book _So lernt man lernen_.
The rule: five boxes, doubling gaps
Picture five boxes. Every card lives in one of them, and each box has a waiting time:
| Box | Next review after |
|---|---|
| 1 | 1 day |
| 2 | 2 days |
| 3 | 4 days |
| 4 | 8 days |
| 5 | 16 days |
Three rules move the cards:
- A new card starts in box 1 and is due straight away.
- If you knew the answer, the card moves up one box and comes back after that box's gap. A card already in box 5 stays there and returns every 16 days.
- If you did not know it, the card goes back to box 1 and comes back tomorrow, however high it had climbed.
The gaps are counted from the day you actually review. Leitner's original boxes were compartments of growing size rather than day counts; fixed doubling days are a common way to put his idea on a calendar, and they are what the tool uses.
A worked example in real dates
Take one card, "capital of Laos → Vientiane", first studied on Wednesday 14 October 2026.
| Review date | Answer | New box | Next due |
|---|---|---|---|
| Wed 14 Oct 2026 | knew it | 2 | Fri 16 Oct |
| Fri 16 Oct | knew it | 3 | Tue 20 Oct |
| Tue 20 Oct | didn't know | 1 | Wed 21 Oct |
| Wed 21 Oct | knew it | 2 | Fri 23 Oct |
| Fri 23 Oct | knew it | 3 | Tue 27 Oct |
| Tue 27 Oct | knew it | 4 | Wed 4 Nov |
| Wed 4 Nov | knew it | 5 | Fri 20 Nov |
| Fri 20 Nov | knew it | 5 | Sun 6 Dec |
Two things stand out. First, the miss on 20 October cost very little in time: the card was back the next day and in box 3 again within a week. Second, after five weeks the card is seen only twice a month. That is the whole point: your time goes to the cards that need it.
A card you know every single time is reviewed on day 0, 2, 6, 14 and 30, then every 16 days (46, 62, 78 and so on). Five reviews in the first month is a modest price for a fact you will still know at the exam.
How much reviewing a day?
This is the question people forget to ask before they import 2,000 words. Suppose you add 10 new cards every day and, to keep the arithmetic clean, answer every card correctly. The number of cards due on a given day, simulated with the same rules the tool uses, is:
| Day | Cards due (10 of them new) |
|---|---|
| 0 | 10 |
| 2 | 20 |
| 6 | 30 |
| 14 | 40 |
| 30 | 50 |
| 46 | 60 |
The load grows in steps: each time a batch reaches a new box, another 10 cards a day join the queue. With a fixed 16-day top box it keeps growing slowly, by about 10 cards every 16 days, because every card you have ever added stays in rotation. Real misses add more, since a missed card restarts at box 1. Two practical conclusions follow. Start with fewer new cards a day than you think, perhaps 5 to 10, and add more only once your daily reviews fit in the time you have. And when a deck is truly learned, archive it by exporting it and starting a fresh one, rather than letting old cards fill every day.
Using the schedule well
- Answer honestly. "Knew it" means you produced the answer before flipping, not that it looked familiar afterwards. Flattering yourself pushes weak cards into long gaps and you meet them again only when they are forgotten.
- Study the due cards, then stop. The queue is short on purpose. Extra cramming the same day does not move cards faster; that is why the tool's Practise all cards mode leaves the schedule alone.
- Lowest box first. The tool orders the queue by box, so the shakiest cards are seen while you are still fresh.
- Skip a day without guilt. Overdue cards wait. Because the next gap is counted from the day you review, coming back on Monday after a lost weekend simply continues the plan.
- Make small cards. One fact per card, a question you can answer in a few seconds. A card that bundles five facts is either always "known" or always "missed", and the box tells you nothing.
Common mistakes
Adding too many new cards at once. Importing a whole vocabulary list on day one means the queue fills with a few hundred box-1 cards and the habit dies in a week. Add a slice each day.
Treating box 5 as "done". A box 5 card still comes back every 16 days. That is cheap maintenance, but it adds up. Archive decks you no longer need.
Reviewing early to "get ahead". Seeing a card before it is due wastes the spacing that makes the review effective, and in a box system it can hide which cards really slipped.
Two directions on one card. If you need "chó → dog" and "dog → chó", make two cards. One card with both directions is never tested properly in either.
Losing the deck. Browser storage is easy to clear. Export a JSON backup from the tool every week or two; it keeps every card's box and due date, so a restore puts you back exactly where you were.
Try it
Open the flashcards tool, make a deck of ten cards and study it today. The progress panel shows how many cards sit in each box and the date of your next review, so you can watch the schedule above play out on your own deck. If you like timed sessions, pair it with the Pomodoro timer.