Flashcard apps quietly decide when to show you each word again. This free tool lets you see how that decision is made — and why the timing makes or breaks your learning.
When you learn a new word, the memory fades over days — that fading is the forgetting curve. Spaced-repetition apps fight it by showing you each word again just before you'd forget it, and every review makes the memory last longer.
Pick a scenario below (or open Customize reviews to build your own) and the chart shows the chance you still remember the word over time. New to this? Start with The Science of Spaced Repetition or the deep dive Why FSRS Is Better Than SM-2.
Tap a review to cycle its grade — Again = forgot it, Hard = barely got it, Good = remembered it, Easy = instant recall. Use +/− to add or remove reviews.
| # | Grade | FSRS | SM-2 |
|---|
In 1885, Hermann Ebbinghaus memorized thousands of nonsense syllables and tested himself at increasing delays. Two findings from that experiment power every flashcard app built since:
Spaced repetition exploits both: review each card just before you'd forget it. The open question is when exactly that moment is — and that's where the algorithms differ.
The schedule and memory state come from the actual FSRS engine (the open-source ts-fsrs library, fitted to hundreds of millions of real flashcard reviews) running on our server — the same code that schedules every real review on Words on Repeat, so what you see here is exactly what the app does. The two schedules differ only in when each algorithm shows you the card:
Dig deeper: FSRS vs SM-2 in detail · the science of spaced repetition.
The forgetting curve, first measured by Hermann Ebbinghaus in 1885, describes how memories fade over time without review. Recall probability drops steeply at first, then more slowly. Each successful review flattens the curve — the memory becomes more stable and you can wait longer before the next review. Spaced repetition systems exploit this by scheduling reviews just before you would forget.
SM-2 (from 1987) multiplies the previous interval by an ease factor that only ever gets punished, so cards drift into "ease hell" and reviews land far from the optimal moment. FSRS models memory directly — it tracks stability and difficulty per card and schedules each review at the exact point where your recall probability drops to a chosen target (typically 90%). In benchmarks on hundreds of millions of real reviews, FSRS predicts recall substantially more accurately than SM-2, which means fewer reviews for the same retention.
90% is the standard default and the best balance for most learners. Raising the target to 95%+ makes intervals much shorter and can double your daily workload for a small retention gain. Dropping below 85% saves reviews but lapses become frequent enough to feel discouraging. Try moving the slider in the simulator to see how strongly the target affects the schedule.
Every review on Words on Repeat is scheduled by FSRS. 185+ free vocabulary decks in 12 languages.
Privacy · Terms · GitHub · Cookie Settings