Leitner System Explained [From Shoebox to 117,812 Cards]

The most copied review algorithm in the world isn’t software. It’s a cardboard box from 1972, divided into five slots with a ruler.

The Leitner system schedules flashcard review by moving cards between numbered boxes. Answer a card correctly and it climbs one box, earning a longer wait before you see it again. Miss it and it falls back for frequent drilling. Every serious spaced repetition tool since, LearnClash included, descends from that box.

This guide covers the 1972 original, a paper setup you can copy tonight, the reason most shoebox systems end up in the attic, and a census of 117,812 live cards inside LearnClash’s three boxes. If you’d rather feel the boxes work before reading about them, duel me on study techniques and come back in three minutes.

What Is the Leitner System?

The Leitner system is a flashcard method that sorts cards into boxes by how well you know them. New and missed cards sit in Box 1 for near-daily review. Each correct answer promotes a card one box, stretching the gap before its next appearance. LearnClash runs the same ladder digitally, with three stages instead of five boxes.

The whole algorithm in one picture. Cards climb on correct answers, fall on misses, and your daily effort pools in the low boxes.

Two research-backed effects do the heavy lifting. The first is retrieval practice: pulling an answer from memory strengthens it far more than rereading it, the well-documented testing effect. The second is the spacing effect: a review lands hardest when it arrives after a delay, right before you’d forget.

The boxes bundle both effects into furniture. You can’t reread a flashcard face-down, so every card is a retrieval attempt. And the box number is the delay. No timer, no app, no schedule to compute.

Box	What it holds	How often you see it
Box 1	New cards and every recent miss	Daily
Middle boxes	Cards you’ve answered correctly once or twice	Every few days to weekly
Last box	Cards you’ve proven repeatedly	Monthly, or retired entirely

The design has one property modern apps still copy: your time pools where your memory is weakest. A card you keep missing keeps returning. A card you know cold nearly stops costing you minutes.

The box count itself is a tuning knob, not doctrine. Leitner used five. Most paper guides teach three. LearnClash ships three stages with deliberately long waits between them. What every working version shares is the sorting rule, because the sorting rule is the algorithm.

The 1972 Original: Five Boxes, Measured in Centimeters

Sebastian Leitner was a German science journalist, and the system he published in his 1972 book So lernt man lernen (“How to Learn to Learn”) was a physical object: a case with five compartments of increasing width. That case is the direct ancestor of the three-stage ladder LearnClash now runs in code.

The 1972 original. Compartment widths of 1, 2, 5, 8, and 14 cm meant capacity, and capacity meant schedule.

The compartment widths were 1, 2, 5, 8, and 14 centimeters. And those widths weren’t packaging. They were the algorithm.

Here’s the detail almost every modern summary drops: Leitner’s original had no calendar. You didn’t review Box 2 “every three days.” You reviewed a box when it was full. The spacing emerged from geometry:

Compartment	Width	Fills up	Effective rhythm
Box 1	1 cm	Within a day or two	Near-daily
Box 2	2 cm	Days	Every few days
Box 3	5 cm	A week or more	Weekly-ish
Boxes 4-5	8 and 14 cm	Weeks to months	Rare

Think about it this way.

A fat box fills slowly, so its cards wait longer. A thin box overflows daily, so its cards return fast. Leitner encoded a review schedule into cardboard dimensions a decade before anyone would code one into software.

Did you know? So lernt man lernen still has no widely available English edition. The most influential study method of the flashcard era spread mostly through summaries, forum posts, and the software that borrowed its boxes.

The book sold steadily in German-speaking Europe for decades, and the box outlived it. When Piotr Wozniak built SuperMemo’s algorithms in the late 1980s, and when Anki brought per-card scheduling to everyone in 2006, both were answering a question Leitner’s cardboard had posed: what if the box widths could adjust themselves per card?

Three Boxes or Twelve: The Variants That Followed

Leitner’s five compartments spawned a family of adaptations, from a stripped three-box weekly rotation to a twelve-deck setup for certification grinders. LearnClash’s three-stage ladder belongs to this family too, on the fewer-boxes-longer-waits end. Picking a variant means picking how much bookkeeping you’ll tolerate.

Four generations of the same sorting rule. More boxes buy finer spacing at the cost of more bookkeeping.

The three-box version is the one most study guides teach, and the one worth starting with. It swaps capacity triggers for a calendar (Box 1 daily, Box 2 every third day, Box 3 weekly) and cuts the promotion ladder to two hops. Less granular than the original, and far more likely to survive contact with a real week.

At the other extreme sits the proficiency variant: ten numbered decks plus a Current deck and a Retired deck, run across ten numbered sessions. A card retires only when it comes up for review during the session whose number matches its deck. It offers the finest-grained spacing a shoebox can deliver, and a rulebook you have to genuinely study first.

Variant	Boxes	Schedule source	Best fit
Leitner 1972 original	5	Box fullness	Purists with one big deck
Three-box	3	Calendar (1/3/7-day rhythm)	Anyone starting on paper
Proficiency (12-deck)	12	Session numbers 0-9	Long exam campaigns
LearnClash 3-stage SRS	3	Per-card timers (7 and 90 days)	People who’d rather play than schedule

Digital tools took the fourth path: dissolve the boxes entirely into per-card timers. SuperMemo and Anki compute a fresh interval for every single card, which is the logical endpoint of Leitner’s geometry. LearnClash went the other way on purpose, keeping three fat boxes and making each promotion mean more. Fewer, harder boxes produce a ladder a player can actually feel climbing mid-game.

The Rule Everyone Softens: Back to Box 1

Leitner’s original demotion rule is brutal. Miss a card, any card, and it returns to Box 1. A card that survived months of climbing to the fifth box gets erased to zero by one blank moment. LearnClash deliberately rejects that rule, and so does nearly every modern implementation, each in its own way.

One miss, two philosophies. The original resets a card to zero; the softened variant docks it one step.

The reset made sense for Leitner’s material. He wrote with vocabulary drilling in mind, and a fully blanked word pair is close to a total loss. But for most knowledge, a miss is rarely total. You blanked on the year of the Berlin Airlift; you didn’t forget the Cold War.

So the variants appeared. The common three-box adaptations demote a card one box instead of resetting it. Anki’s lapse handling is its own negotiation with the same problem, closer to the harsh original than most users realize. The full comparison lives in the LearnClash retention-curve breakdown, which walks through what each tool does to a lapsed card.

Rule	Missed card goes to	One bad day costs
Leitner 1972 original	Box 1, always	Everything the card had earned
Common three-box variant	Down one box	One promotion
LearnClash 3-stage SRS	Down exactly one stage	One promotion, never more

LearnClash’s version is written into the answer-handling code as a single rule: a wrong answer drops the card exactly one stage. A Mastered card that lapses becomes Known, waits out a fresh 90-day gap, and gets its chance to re-prove itself. It does not return to the daily grinder.

Our position, and we hold it firmly: the full reset punishes a bad day like a bad month, and it’s the first rule any sane implementation softens. Memory fades by degrees. The demotion should too.

How to Run a Paper Leitner System

You need index cards, three boxes, and a pen. LearnClash automates this exact loop with server-side timers, but the paper version is the fastest way to understand what any SRS is doing under the hood, and it works with zero screens involved.

The three-box week. Box 1 daily, Box 2 every third day, Box 3 weekly. Graduating out of Box 3 retires the card.

The three-box version runs on a calendar instead of box widths, and it goes like this:

1. Write your cards. One question per card, answer on the back. Keep the batch small; 20 to 40 new cards is plenty to start.
2. Everything begins in Box 1. No exceptions, including material you think you know.
3. Review Box 1 daily. Read the question, answer out loud before flipping, then grade yourself. Correct cards move to Box 2. Missed cards stay.
4. Review Box 2 every third day. Correct cards graduate to Box 3. Missed cards drop back to Box 1 (or one box, if you’ve adopted the gentler rule; we would).
5. Review Box 3 weekly. Correct cards retire from the rotation. Missed cards fall back.
6. Feed new cards into Box 1 only as the box empties. A starving Box 1 is a healthy Box 1.

Here’s the first week with a 20-card deck of European capitals, so the flow is concrete:

Day	Session	What happens
Day 1	Box 1 (all 20 cards)	You get 13 right; they climb to Box 2. Seven stay.
Day 2	Box 1 (the 7 holdouts)	Four climb. Three stay for tomorrow.
Day 3	First Box 2 session (13 cards)	Eleven graduate to Box 3. Two fall back down.
Day 7	First Box 3 review	Anything you nail here leaves the rotation for good.

By day seven, a deck that cost 20 reviews on Monday costs a handful, and that shrinking daily bill, more than any single test score, is the visible proof that the sorting rule is doing its job.

Two ground rules keep it honest. Grade “sort of knew it” as a miss; hesitation on paper becomes a blank on the exam. And never skip a scheduled day to “save up” reviews, because the backlog compounds. Which brings us to how these boxes die.

Why Paper Leitner Systems Die: Review Debt

The paper Leitner system has a structural enemy: Box 1 fills from two directions. Every new card starts there, and every miss from every other box lands back there. Skip three days and the daily box becomes an hour-long wall of cardboard. LearnClash’s census data, further down, shows the same debt pattern surviving into software.

Review debt in one image. New cards and failed cards both feed Box 1, and skipped days multiply it.

The debt spiral has a predictable script:

1. A missed Tuesday doubles Wednesday’s stack.
2. The bigger stack feels worse, so Wednesday slips too.
3. Within two weeks the box stops being a schedule and becomes a guilt object on a shelf.

And the failure isn’t laziness. It’s that the system’s cost is front-loaded exactly where motivation is thinnest, on the material you keep getting wrong.

Debt is one of three killers, and the other two are quieter:

Killer	What it corrodes
Review debt	The schedule itself
Self-grading drift	The accuracy of every box
The fluency illusion	Your will to continue

Self-grading drift works like this. On paper, you are both student and referee, and the referee softens. “Close enough” becomes correct, cards climb boxes they haven’t earned, and the schedule quietly stops matching your actual memory. A quiz with server-graded answers can’t be sweet-talked, which is one unglamorous reason digital systems keep their accuracy.

The fluency illusion is crueler: it makes people abandon the method precisely when it’s working. Nate Kornell’s 2009 flashcard experiments had students study GRE-type word pairs either as one large spaced stack or as four small stacks drilled one at a time. Spacing won for 90% of participants. Yet after the first session, 72% believed the massed stacks had worked better. The method that wins feels like the one that’s failing, so learners quit it.

“Across experiments, spacing was more effective than massing for 90% of the participants.” (Kornell, Applied Cognitive Psychology, 2009)

Paper Leitner asks you to override that feeling with discipline, every day, indefinitely. Some people manage it. The census below shows what happens when the discipline requirement is removed instead.

117,812 Cards Later: A Census of Three Live Boxes

LearnClash’s July 2026 census counted 117,812 live cards across 1,022 players: 48,300 in the Learning box (41.0%), 67,476 in Known (57.3%), and 2,036 Mastered (1.7%). Each card is one player’s history with one question, sorted by the same climb-and-fall rules Leitner drew in 1972.

The three boxes at production scale. Most of the corpus sits in the middle box, waiting out its 90-day proving period.

Method: read-only export of the LearnClash production database on July 15, 2026, covering every per-player card record (117,812 cards, 1,022 players), with box counts cross-checked against a second, independently maintained per-topic ledger. The two ledgers agree within 0.3%.

The results surprised us.

LearnClash box	Cards	Share	Wait before next review
Learning (Box 1)	48,300	41.0%	7 days
Known (Box 2)	67,476	57.3%	90 days
Mastered (Box 3)	2,036	1.7%	Never again; card retires

That 1.7% mastery rate looks harsh until you see the gate it guards. Reaching Mastered requires a correct answer, then a second correct answer after the full 90-day wait. There is no faster path; answering early during the wait doesn’t advance the card. Mastery in LearnClash means you held a fact across three months, so the third box fills at the speed of proof, and no faster.

Zoom into the waits themselves and the census gets more textured:

Card state on census day	Cards	Timing detail
Inside the 7-day Learning wait	3,783	Median 4.4 days from its next review
Inside the 90-day Known wait	50,408	Median 38.9 days from its next review
Past due for review	61,585	Median overdue Learning card: 55 days late

That bottom row is the familiar one: review debt survives digitization. More than half the corpus was due on census day. Players chase new topics; old misses pile up.

It’s Box 1 overflowing, at production scale.

Counting caveat: the demotion rule leaves no per-event log (it’s a single line of answer-handling code), so an honest census counts stocks, not falls. One such stock: 2,175 Learning-box cards in the LearnClash census had been answered again in a later session and still sat in Learning, cards that got knocked back down or never escaped.

But here’s where it gets interesting. In LearnClash the debt clears through play rather than discipline, because the question picker treats due cards as first-class candidates: when you start a duel or practice session, cards past their review date compete with brand-new questions for the slots. The overflowing box empties as a side effect of wanting to play.

One more census number tells that story. Of first encounters with a card, 93.8% happened inside 1v1 duels (111,014 duel answers against 7,361 in solo practice). And across all 118,375 first encounters, players got 56.9% right on the first try. The boxes get filled, and refilled, by people trying to win a match, with the Leitner machinery running silently underneath.

How LearnClash Modernized the Leitner System

LearnClash’s 3-stage SRS is a Leitner box rebuilt for a quiz duel: three stages instead of five compartments, per-card timers instead of full-box triggers, and a competitive game doing the daily-discipline job. The sorting rules would be immediately recognizable to a 1972 reader.

Same ladder, new machinery. Every 1972 mechanism has a direct 2026 counterpart.

The translation, piece by piece:

Leitner 1972	LearnClash 2026
Five cardboard compartments	Three stages: Learning, Known, Mastered
Review when a box is full	Per-card timers: 7 days, then 90
Missed card resets to Box 1	Missed card drops exactly one stage
Cards retire from the last box	Mastered cards leave the schedule, stay playable
You grade yourself	The server grades a timed quiz answer
Daily discipline required	Duels re-serve due cards as part of play

Three of those rows carry most of the philosophy. The one-stage demotion treats memory as fading by degrees, as covered above. The 90-day proving period makes the last box mean something; it’s why only 1.7% of cards have cleared it. And the duel delivery attacks the real reason shoebox systems fail, which was never the spacing math. It was showing up.

Key takeaway: every 1972 mechanism survives in the 2026 build. What changed is who pays the discipline cost: Leitner billed the learner daily, LearnClash bills the game.

So the fluency illusion from Kornell’s experiments gets handled structurally too, because you never have to believe spacing is working or even think about the calendar: the 1/3/7/21 folklore schedule that circulates on study blogs asks you to run the bookkeeping by hand, while a duel deals you the card the schedule already chose. Sebastian Leitner solved scheduling with a ruler. Fifty-four years later the boxes are server-side, the referee is incorruptible, and the daily discipline is a game you’d play anyway; the learning-science hub collects the rest of the machinery behind that claim.

The box was never the point. The schedule was. Keep the schedule, lose the cardboard, and let the game do the sorting.