Does the Leitner System Actually Work

Does the Leitner system work? Mostly yes, though the research supports something slightly different from the box on your desk. The idea underneath it has good evidence behind it. That idea is spaced repetition: reviewing a fact at growing intervals and pulling the answer from memory each time. Controlled studies of flashcards on spaced schedules show real gains in how much students remember weeks later.

What the evidence does not cover is Sebastian Leitner's own recipe. He published it in his 1972 book So lernt man lernen (How to Learn How to Learn), and no source here tests its exact box intervals against alternatives. Nearly all of the evidence also measures recall of facts. Applying those facts to new problems is far less tested. The method works when you run it honestly. It can also tell you that you know something when you only recognize it.

Every card starts in Box 1, which you review most often. Answer a card correctly and it moves up one box, where you see it less often. Miss it and it drops all the way back to Box 1, whichever box it was in. One setup guide calls that full demotion "a key feature of the system".

The result is that you spend most of your time on cards you don't know yet, and less and less on the ones you do. Each correct answer earns a longer gap before the next review.

Blogs disagree on the gaps themselves. One gives five boxes at 1, 2, 4, 9 and 14 days. Another uses 1, 2, 7, 14 and 30. A third lists 1, 3, 5, 8, 16 and 31. That spread tells you something. No single schedule has been shown to be the correct one, so each source picked a sensible-looking set of numbers.

None of these studies used physical boxes. All of them used flashcards on spaced schedules, which is the mechanism the boxes are built to deliver.

Paediatrics students, 2025. In a quasi-experimental study in Frontiers in Medicine, 90 final-year medical students in India were split into two groups. One group studied with digital flashcards reviewed at 1, 3, 7, 14 and 28 days, a Leitner-like progression. The other used textbooks and clinical notes. The groups started level, with pre-test scores of 11.42 and 11.58 out of 20. The flashcard group finished at 16.24, while the textbook group finished at 11.89 (p<0.0001). The authors list the limits themselves: there was no long-term follow-up, and students knew they were in the intervention, which may have changed how hard they worked. Also, 17 of the 20 test questions were recall questions and only 3 asked students to apply what they knew.

Dental students, 2024. A randomized controlled trial in the Journal of Dental Education compared 45 students who used a spaced-repetition flashcard app after a lecture with 45 who had the lecture only. Right after the session, the two groups scored about the same. At one month and again at three months, the flashcard group scored significantly higher (p≤0.001 at both points). This is the pattern to expect from spacing. It does little for tomorrow's quiz and a lot for whether you still know the material in a few months.

Microbiology students, 2024. At UNC Chapel Hill, an introductory course added instructor-made Anki cards in one semester and compared the results with a semester that had none. Roughly 140 to 150 students were in each. Overall exam scores did not improve significantly (p=0.2657). Only the first of three exams rose, from 70.55% to 74.87%. Students did feel more confident. The authors noted that these better perceptions "are not contingent on performing significantly better on assessments."

When a blog says the Leitner system is "backed by research," it usually means research on spacing in general. That is a fair basis for confidence in the idea. It does not show that Leitner's particular rules are the best way to carry the idea out.

Algorithms like SM-2 and FSRS, which power apps such as Anki, are descendants of the box. They set each card's interval individually. The benchmarks quoted on blogs compare these algorithms with each other, not with Leitner, so none of the figures here measures how much efficiency a paper box gives up.

The biggest risk comes from grading yourself. You flip a card, see the answer, and think "right, I knew that." But you didn't produce the answer. You recognized it. The card moves up a box anyway, and the system now treats a fact you can't recall as nearly learned. One guide says the method works best when "you are honest about self-grading." In practice that means you say or write the answer before you flip, and you count the card as correct only if what you produced matches.

The other failures are mechanical:

• Skipping demotion. Moving a missed card back only one box, or not at all, removes the step that focuses your time on weak cards.
• Reviewing every box every day. That removes the spacing completely.
• Flooding Box 1. One FAQ calls adding too many new cards at once the most common reason Leitner systems collapse, and suggests 5 to 10 new cards a day.
• Testing the wrong thing. A card that asks you to recall a definition trains you to recall that definition. None of the studies above shows that this helps on problems you haven't seen before.

Use it if you are memorizing a few hundred discrete facts, such as vocabulary, drug names or developmental milestones, and you want a system you can see and handle. The low-tech version costs a shoebox and index cards, and you can build the box at home in an evening. Expect the payoff to show up weeks later, not on the next quiz.

Before you trust it, check two things. Do you produce each answer before you flip the card? And does your exam ask you to recall facts or to apply them? If it asks you to apply them, flashcards can cover the facts, but you'll still need practice problems to learn how to use them.

If your deck grows into the thousands, the sources here agree that software handling per-card timing is the better fit. Whichever exact intervals you choose matter less than reviewing on the days your schedule says to.