For medical school, FSRS beats SM-2: it schedules each card by predicted retrievability instead of fixed ease multipliers, so cards stop getting trapped at the 130% ease floor. Switch FSRS on in Anki 23.10 or later, set Desired Retention to 0.88-0.90 for dedicated study (0.85-0.87 on rotations), optimize after a few hundred reviews, and simulate your load before raising the target.
What is FSRS and Why Does It Replace SM-2?
For more than fifteen years, Anki scheduled flashcards using SM-2, an algorithm published by Piotr Wozniak in 1987. While pioneering, SM-2 exhibits structural mathematical limitations when handling large, 30,000-card medical school decks like AnKing:
- Ease Hell: In SM-2, pressing "Hard" subtracts 0.15 EF units (15 percentage points) from a card's Ease Factor. Over months of clinical study, hundreds of cards get trapped at the 130% ease floor (see our protocol on how to fix Anki Ease Hell), creating an unmanageable review treadmill.
- Rigid Heuristic Multipliers: SM-2 assumes uniform memory decay rates across cards regardless of cognitive complexity or individual student retention patterns.
FSRS (Free Spaced Repetition Scheduler) replaces fixed multipliers with a three-component model of human memory: Difficulty (D), Stability (S), and Retrievability (R) (the DSR model). Developed by the Open Spaced Repetition project and integrated natively into Anki 23.10+, FSRS optimizes interval growth based on your personal review history. Complete documentation is maintained in the official Anki Manual on FSRS.
The Core Parameter: Desired Retention
Under SM-2, students had no direct way to set their target retention rate; they adjusted "interval modifiers" heuristically. FSRS introduces direct control via Desired Retention (DR):
- 0.90 (90% Retention): The standard target for dedicated USMLE Step 1 and Step 2 CK study. It maintains strong memory recall while keeping daily review queues manageable.
- 0.85 – 0.87 (85% – 87% Retention): Recommended during third-year clinical rotations (e.g., Surgery, Inpatient Medicine). Lowering target retention from 0.90 to 0.85 materially reduces daily review load (how much depends on the collection; simulate your own numbers with the FSRS calculator linked below), liberating critical hours for patient care and question bank practice.
- 0.95+ (The Diminishing Returns Trap): Escalating target retention from 0.90 to 0.95 sharply increases review frequency: the Anki Manual warns workload rises very quickly above 90%, and per-card review counts more than double going from 90% to 95% for a marginal 5% gain in factual recall. Run your own collection through the FSRS calculator linked below before raising your target, since daily load depends on collection size and review history.
Recommended FSRS Settings for Medical School
| Parameter | Recommended Value | Mathematical Rationale |
|---|---|---|
| Desired Retention | 0.88 – 0.90 | Optimal balance between retention fidelity and daily review volume. |
| Learning Steps | 10m or 15m | Single intra-day step. FSRS calculates first-day interval dynamically from card stability. |
| Relearning Steps | 10m | Single intra-day step. Avoid multi-day relearning steps that disrupt FSRS lapse tracking. |
| Maximum Interval | 36,500 days (Default) | Keep Anki's default (100 years). Artificially capping intervals at 180–365 days forces mature, consolidated cards back into redundant daily review loops without cognitive benefit. |
Rating Button Mechanics Under FSRS
Under FSRS, rating buttons feed directly into personalized weight optimization. Misusing buttons degrades scheduling accuracy:
- Again (1): Complete recall failure. The card was forgotten and must be relearned.
- Hard (2): Successful recall requiring noticeable hesitation or significant cognitive effort. In FSRS, pressing Hard registers a successful pass with elevated difficulty, calculating a moderate interval expansion rather than an SM-2 ease penalty.
- Good (3): Standard successful recall with normal effort. Under the two-button paradigm, Good serves as your default pass button.
- Easy (4): Instantaneous, effortless recall. Schedules an aggressive interval expansion.
Migration Safety: Preserving Your Review History
When enabling FSRS for an existing collection, keep Reschedule cards on change toggled OFF. This prevents FSRS from instantly shifting thousands of future review dates across your deck. Anki evaluates cards under FSRS progressively as they naturally come due. If you ever need to revert deck settings, Anki provides an instant safety net via Edit → Undo Deck Options (Ctrl+Z / Cmd+Z immediately after saving).
Optimizing FSRS Parameters
Once you have a few hundred reviews in your collection, scroll down to the "FSRS parameters" section in Deck Options and click Optimize (current Anki versions optimize with whatever history exists; only older versions required 1,000). The machine learning optimizer will calibrate 21 parameters (FSRS-6) to match your personal memory decay curve across each clinical deck preset.
Simulate Your Review Workload
Unsure whether shifting your retention target will overload your schedule? Use the Anki FSRS Target Retention Calculator to simulate review volume, or upload your collection directly to the Penziv Deck Analyzer to verify your True Retention across every clinical subject.
If you are returning from an exam block or intense clinical shift with hundreds of overdue cards already accumulated, review our guide on how to clear an Anki backlog using relative overdueness and FSRS Postpone so you can safely recover without interval distortion.
Frequently Asked Questions
Should I switch from SM-2 to FSRS?
Yes for a medical school deck: FSRS is native in Anki 23.10+, models each card's memory instead of applying one ease multiplier, and removes the ease-floor trap. Keep "Reschedule cards on change" off when switching so existing due dates do not jump.
What Desired Retention should medical students use?
0.88-0.90 for dedicated study (Anki's 0.90 default is the balance point), 0.85-0.87 on clinical rotations when time is short. Above 90% workload rises very quickly, so raise the target only for a short high-stakes window, never as a permanent setting.
How many reviews do I need before clicking Optimize?
A few hundred. Current Anki versions optimize with whatever history exists; re-optimize about monthly as more history accumulates.
Does switching to FSRS fix my existing ease-hell cards?
It stops the mechanism, it does not rewrite history: FSRS has no ease factor, so no card can get pinned at a floor again, but cards already trapped need the ease-hell protocol and backlog triage first.
Simulate your retention target
Simulate your retention target
Calculate your daily review volume across retention settings before changing options, and evaluate your collection to verify your True Retention.