If you've been using Anki for a while, you've probably heard about FSRS — the Free Spaced Repetition Scheduler. It replaced Anki's decades-old SM-2 algorithm as the default in Anki 23.10, and it's a genuine leap forward. Cards get scheduled more accurately, you do fewer unnecessary reviews, and the whole system adapts to how your brain actually works.
But there's a lot of confusion about what FSRS actually does, how to set it up properly, and what the settings mean. This guide covers everything: what FSRS is, how to enable and configure it, how to optimize your parameters, and how to get the most out of it.
What Is FSRS?
FSRS (Free Spaced Repetition Scheduler) is a modern spaced repetition algorithm developed by Jarrett Ye. It uses a mathematical model of memory — based on research into how humans forget — to predict exactly when you'll forget each card. Then it schedules your review right before that happens.
The key difference from the old SM-2 algorithm: FSRS actually models your memory. SM-2 uses simple multipliers (your "ease factor") that go up when you get cards right and down when you get them wrong. It doesn't really understand forgetting — it just adjusts intervals mechanically. FSRS uses a three-component model of memory that tracks:
- Stability: How long you can go before forgetting (measured in days)
- Difficulty: How inherently hard the card is for you
- Retrievability: The probability you can recall the card right now
This means FSRS can make much better scheduling decisions. Easy cards get pushed out further, hard cards get reviewed more often, and the whole system tunes itself to your personal learning patterns.
Why Switch to FSRS?
Fewer Reviews, Same Retention
The biggest practical benefit: FSRS typically reduces your daily review count by 20-30% compared to SM-2 while maintaining the same retention rate. Some users report even bigger reductions. This happens because SM-2 tends to over-schedule easy cards and under-schedule hard ones. FSRS gets the timing right for both.
Personalized Scheduling
FSRS learns from your review history. When you optimize your parameters (more on this below), the algorithm adjusts to your personal forgetting curves. If you're someone who retains vocabulary easily but struggles with formulas, FSRS will adapt its scheduling accordingly.
Desired Retention Control
With FSRS, you set a "desired retention" rate — the probability you want to recall any given card when it comes up for review. Set it to 0.90 and FSRS schedules cards so you'll remember about 90% of them at review time. This is much more intuitive than SM-2's ease factors and interval modifiers.
Better Handling of Overdue Cards
If you take a break and come back to a mountain of overdue reviews, FSRS handles this gracefully. It considers how long the card has been overdue when calculating the next interval. If you missed a review by a week but still got it right, FSRS recognizes that your memory is stronger than expected and extends the interval accordingly.
How to Enable FSRS
FSRS has been the default scheduler since Anki 23.10, so if you installed Anki recently, you might already be using it. Here's how to check and enable it:
Desktop (Anki 23.10+)
- Open Anki and go to Tools → Preferences
- Click the Scheduling tab (or Review tab in older versions)
- Look for FSRS — make sure it's enabled
- If you're upgrading from SM-2, Anki will ask if you want to switch. Click yes.
AnkiMobile (iOS)
- Go to Settings → Review → Scheduling
- Enable FSRS
AnkiDroid
- Go to Settings → Advanced → Scheduler
- Select FSRS (requires AnkiDroid 2.17+)
Important: FSRS settings sync across devices through AnkiWeb. You only need to configure it once on desktop and it'll apply everywhere.
Configuring FSRS Settings
Once FSRS is enabled, you'll see several new settings in your deck options. Here's what each one means and how to set it.
Desired Retention
This is the most important setting. It controls how often you want to successfully recall cards at review time.
- Default: 0.90 (90% recall rate)
- Range: 0.70 to 0.97
What to set it to:
- 0.85-0.90: Good for most users. Balances review load with retention.
- 0.90-0.95: Use for high-stakes material (board exams, professional certifications). More reviews but stronger retention.
- 0.80-0.85: Use if you're drowning in reviews and willing to accept more "again" presses in exchange for fewer daily cards.
- Above 0.95: Not recommended. Reviews increase exponentially for marginal retention gains.
For medical students studying for Step 1: 0.90 is the sweet spot. For language learning where approximate recall is fine: 0.85 works well.
Maximum Interval
The longest interval FSRS will ever assign to a card.
- Default: 36500 days (100 years — effectively no limit)
- Practical recommendation: 365 days if you're studying for an exam in the next year
If you're preparing for a specific test, set this to the number of days until your exam. This ensures you see every card at least once before test day.
SM-2 Retention
This only matters if you're transitioning from SM-2. It tells FSRS what retention rate SM-2 was achieving so it can properly convert your existing intervals. Leave this at the default unless you have reason to change it.
FSRS Parameters
This is a string of numbers that defines your personal memory model. Don't edit these manually. Use the optimizer instead (see below).
Optimizing Your Parameters
This is where FSRS gets powerful. The optimizer analyzes your review history and generates parameters tuned to your personal forgetting curves.
How to Optimize
- Go to Deck Options
- Scroll to the FSRS section
- Click Optimize FSRS Parameters
- Wait for it to finish (takes a few seconds to a few minutes depending on your collection size)
- Click Save
When to Optimize
- After your first 1,000+ reviews: The optimizer needs data to work with. Less than 1,000 reviews means it's estimating with limited information.
- Every 1-3 months: As you accumulate more reviews, reoptimizing improves accuracy.
- After changing study habits: If you went from studying 30 minutes a day to 2 hours, your forgetting patterns may shift.
Per-Preset Optimization
You can optimize parameters separately for different deck presets. This is useful if you have very different types of material — for example, one preset for vocabulary cards and another for complex concept cards. Each will get parameters tuned to that type of content.
FSRS vs SM-2: Practical Differences
If you're coming from SM-2, here's what changes in practice:
No More Ease Factor
SM-2's ease factor was a multiplier that determined how fast intervals grew. Getting cards wrong tanked your ease factor, creating the dreaded "ease hell" where cards got stuck in short intervals forever. FSRS doesn't use ease factors at all. Difficulty is modeled differently and doesn't suffer from the same death spiral.
Intervals Feel Different
FSRS might give you very short intervals for new hard cards (1 day, 2 days, 4 days) and very long intervals for well-learned easy cards (months). SM-2 tended to be more uniform. Trust the algorithm — it's working from a model of your actual memory.
The "Hard" Button Works Properly
In SM-2, pressing "Hard" was often counterproductive because it hammered your ease factor. In FSRS, "Hard" is genuinely useful — it tells the algorithm you struggled but still recalled the answer, and FSRS adjusts accordingly without permanent damage.
"Good" Is the Default
For most reviews, just press "Good." Only use "Again" when you genuinely couldn't recall the answer. "Easy" should be rare — reserved for cards you recall instantly with zero effort. "Hard" is for cards you recalled but had to really work for.
Common FSRS Questions
Should I Reschedule Existing Cards?
When you switch to FSRS, you'll see an option to "Reschedule cards on change." This recalculates all your existing card intervals using FSRS. Opinions are mixed:
- Yes, reschedule if you want FSRS to fully take over immediately. You might get a spike in reviews for a few days as cards get adjusted.
- No, don't reschedule if you want a gradual transition. FSRS will take over as cards naturally come up for review.
For most users, rescheduling is fine and recommended.
Does FSRS Work for All Card Types?
Yes. FSRS works equally well for basic cards, cloze deletions, and image occlusion cards. The algorithm doesn't care about card format — it models your memory of each individual card regardless of type.
Can I Use FSRS with Filtered Decks?
Yes, FSRS works with filtered decks. However, reviews done in filtered decks may not contribute to the optimizer's data in the same way as regular reviews, depending on your settings.
What About Add-ons?
Most scheduling-related add-ons are unnecessary with FSRS. Add-ons like "ResetEZ" or "Straight Reward" were designed to fix SM-2's problems — problems that FSRS doesn't have. The FSRS4Anki Helper add-on can provide additional statistics and features, but it's optional.
Is FSRS Better for Specific Subjects?
FSRS is better than SM-2 for everything, but the improvement is most noticeable for:
- Large collections (5,000+ cards): FSRS is much better at managing review load across big decks
- Diverse difficulty: Collections with both easy and hard cards benefit most from FSRS's adaptive scheduling
- Long-term study: FSRS's memory model gets more accurate over time
Tips for Getting the Most Out of FSRS
1. Be Honest with Your Ratings
FSRS's accuracy depends on your ratings reflecting your actual memory. Don't press "Good" when you barely remembered — use "Hard." Don't press "Again" because you hesitated for a second — if you recalled it, it's at least "Hard."
2. Study Consistently
FSRS works best when you study daily. Irregular study patterns make it harder for the optimizer to model your forgetting curves accurately. Even 15 minutes every day is better than 2 hours every third day.
3. Don't Fear the "Again" Button
Pressing "Again" doesn't mean failure. It tells FSRS your memory for that card needs reinforcement, and FSRS will schedule it appropriately. With SM-2, pressing "Again" felt punishing. With FSRS, it's just data.
4. Check Your True Retention
Anki shows your actual retention rate in the stats. Compare it to your desired retention. If they're close, FSRS is working well. If your true retention is much higher than desired, you might be over-studying — consider lowering your desired retention to free up time. If it's lower, raise it.
5. Use the "Compute Minimum Recommended Retention" Feature
FSRS can calculate the optimal retention rate that minimizes your total study time. This takes into account both review time and relearning time. The result is usually between 0.85 and 0.90.
Making Cards That Work Well with FSRS
FSRS schedules cards optimally, but it can't fix bad cards. Good card design matters just as much with FSRS as it did with SM-2:
- Keep cards atomic: One fact per card. If a card tests multiple things, FSRS can't model your memory for each piece separately.
- Use cloze deletions: They force active recall and are easy to make from your notes.
- Add context: Include source material, images, or mnemonics. Faster recall means less study time.
- Avoid ambiguity: If a card could have multiple correct answers, you'll rate inconsistently, and FSRS's model will be less accurate.
If you're looking for a fast way to create well-structured Anki cards, SlideToAnki lets you upload lecture slides and automatically generates cloze deletion cards. Each card follows the minimum information principle, and you can review and edit them before exporting to Anki. It's particularly useful for students who want quality cards without spending hours on card creation.
The Bottom Line
FSRS is a meaningful upgrade over SM-2. It schedules more accurately, requires fewer reviews, adapts to your learning patterns, and eliminates long-standing problems like ease hell. If you're still on SM-2, switch. If you're already on FSRS, optimize your parameters and adjust your desired retention.
The algorithm handles the scheduling. Your job is to make good cards, study consistently, and rate honestly. Everything else is automatic.