Why does Anki show you a card after 1 day, then 4 days, then 10 days? Here's the science behind spaced repetition intervals - and how to know if yours are optimized.
Memory decays along a forgetting curve. The optimal time to review is just before you forget - not too early (wasting a review), not too late (after forgetting). Spaced repetition intervals are calculated to hit that precise moment, and they grow longer each time you successfully recall.
Each time you successfully recall a memory, you don't just remember it - you strengthen it. The memory becomes more stable. A more stable memory decays more slowly. This is why the intervals grow exponentially:
| Review # | Typical interval | What happened |
|---|---|---|
| Learning (1st) | 10 minutes | New card, very fragile memory |
| Graduating | 1 day | Card leaves learning queue |
| Review 1 | 3–4 days | First long-term review |
| Review 2 | 8–12 days | Stability increasing |
| Review 3 | 20–35 days | Memory becoming robust |
| Review 4 | 2–4 months | Long-term memory |
| Review 5+ | 6 months – 2 years | Near-permanent memory |
These are approximate. Your actual intervals depend on which buttons you press and the algorithm used (SM-2 vs FSRS). FSRS produces more accurate intervals by modelling each card individually.
When you press Again (forgot the card), two things happen:
This is why good card design matters so much. A card that you consistently fail because it's ambiguous or tests two things at once will never have a long interval - it'll perpetually reset, wasting review time. See how to create Anki cards for how to fix this.
In Anki's old SM-2 algorithm, every Again press reduces the card's "ease factor" by 20%. Ease factor determines how fast intervals grow. After several Again presses, a card might have an ease factor of 130% - meaning it barely grows at all, and you'll be seeing it every few days forever.
This is called "ease hell." Cards stuck in ease hell consume a disproportionate amount of your daily reviews. The fix is either:
FSRS replaces the ease factor with a stability variable. Stability measures how long it will take for your recall probability to drop to your target retention (e.g., 90%). The next interval is simply calculated as the time until you're expected to drop below 90% recall.
This means:
The most common mistake is obsessing over whether your intervals "look right." If you're using FSRS with a 90% retention target, your intervals are already optimized for your personal review history. Trust the algorithm.
What actually matters: reviewing every day (consistency beats optimization), having well-designed atomic cards, and not pressing Easy to inflate intervals artificially. See the Anki settings guide and FSRS algorithm explanation for the full setup.
One of the most misunderstood aspects of spaced repetition is the "Desired Retention" setting. Many users instinctively aim for 95% or 99% retention, believing that the more they remember, the better. However, the relationship between retention and workload is not linear—it is exponential. To move from 90% retention to 95%, you don't just add a few more reviews; you may nearly double your daily workload.
This happens because as you push for near-perfect recall, the algorithm must shorten the intervals significantly to prevent that tiny 5% margin of error. For most learners, a target of 85% to 90% is the "sweet spot." This allows intervals to grow much faster, reducing burnout while still maintaining a high level of mastery. When choosing your target, consider these factors:
A common frustration is the "leech" card—a card that you consistently fail regardless of the algorithm. This is usually not a failure of the spaced repetition interval, but a failure of encoding. Spaced repetition is a tool for retention, not acquisition. If you attempt to memorize a fact you don't actually understand, you are relying on rote memorization, which is far more fragile and decays much faster than conceptual memory.
To fix this, you must ensure the information is "pre-processed" before it enters your SRS. This means breaking complex ideas into atomic pieces and ensuring you can explain the concept in your own words first. Tools like StudyCards AI can help bridge this gap by transforming dense material into optimized, atomic cards that follow the Minimum Information Principle, making them significantly easier for the brain to stabilize.
Proper intervals only matter if your cards are well-designed. StudyCards AI creates atomic, context-rich flashcards from your notes - structured to work perfectly with Anki's spaced repetition system.
Start Free - Build Your Optimized Deck →Generate Anki flashcards free