Blog

The Anki Settings Medical Students Actually Need (FSRS Included)

By , MD, FRCPC · Published November 17, 2026 · Last updated November 8, 2026 · 14 min read

A single stylised student at a desk turning one large dial, with a tall stack of cards on one side shrinking as the dial

Disclosure: Alma publishes this article and makes one of the study tools it discusses.

Short answer: for most medical students, five settings do nearly all the work. Turn on FSRS, leave desired retention at the default 90% until you have a reason to change it, size your new cards per day to the hours you really have, put a ceiling on maximum reviews per day, and keep learning steps shorter than one day. Everything else is polish.[1]

This guide is written for US medical students (M1 and M2 with a lecture-driven deck, dedicated Step prep, and M3 on rotations). Every setting name and default below was checked against the Anki manual on November 7, 2026, when the current desktop version listed on Anki's download page was 26.09.4.[1][2] Where a number is our own starting point rather than something Anki or a study says, we label it as such.

Anki desktop is free and works well. Alma, the company that publishes this article, makes a flashcard app too, and a short note on it sits near the end. This page is about getting the most out of Anki either way.

What are the Anki settings that matter most?

Five settings change your daily workload and memory the most: FSRS on or off, desired retention, new cards per day, maximum reviews per day, and (re)learning steps. Anki's manual says to spend a few weeks on the defaults before adjusting, because a mistaken change can reduce Anki's effectiveness.[1]

Setting (exact name in Deck Options) What it does, per the manual Where to set it
FSRS (bottom of Deck Options) Replaces the older SM-2 scheduler; one switch for the whole collection, not per preset Global
Desired retention The share of cards you should remember when they come due; default 90% Per preset (newer versions also allow per deck)
New cards/day How many new cards are introduced each day you open Anki Per preset, per deck, or "Today only"
Maximum reviews/day A ceiling on review cards shown in a day; the rest wait Per preset, per deck, or "Today only"
Learning steps and Relearning steps The short delays before a new or lapsed card becomes a normal review Per preset
Easy Days Shifts due dates a little to lighten chosen days of the week Per preset

How do I turn on FSRS?

Open Deck Options, scroll to the FSRS section and switch it on. Anki's own short guide is four steps: enable FSRS, make every learning and relearning step shorter than one day, click Optimize under FSRS parameters, then choose a desired retention. FSRS can only be enabled globally.[1]

Before you start:

  1. Check every device. The manual says Anki 23.10, AnkiMobile 23.10 and AnkiWeb support FSRS, and AnkiDroid supports it from 2.17. If one of your devices does not, scheduling breaks.[1]
  2. Fix your steps. Steps such as 10m or 30m are good. The manual recommends steps shorter than 1d and says even 23h is not recommended, because you cannot finish that step on the same day as the first review.[1]
  3. Click Optimize under FSRS parameters. If Anki says the parameters currently appear to be optimal, that is fine.[1]
  4. Leave "Reschedule cards on change" off when you first switch from SM-2. The manual does not recommend it then, because it can make a large number of cards due at once. If you want it anyway, take a backup first.[1]
  5. Do not copy parameters from someone else or edit them by hand; the manual says to let the optimizer set them, and that optimizing once a month is enough.[1]

One habit matters more under FSRS than before. FSRS treats Hard as a pass. If you press Hard when you actually forgot the answer, your intervals will come out unreasonably long. Press Again when you forget.[1]

What desired retention should I pick?

Leave it at 90% unless you have a clear reason. Desired retention is the single most important FSRS setting, and the manual says higher retention means shorter intervals and more reviews per day. Above 90% the workload climbs quickly, and above 97% it can be overwhelming.[1]

The trade-off is a time budget, not a quality score. At lower values you review much less and forget more; as the value approaches 100% the workload rises drastically.[1] In current versions the "Help Me Decide" simulator (marked experimental in the manual) estimates your own workload at different retention levels, and it suggests two questions: how many reviews can I realistically do per day, and do I want to study this much more to remember this much more.[1]

A note on older advice: the manual says the "compute minimum recommended retention" button was removed as of Anki 25.07. Guides that tell you to click it were written for an earlier version.[1]

How many new cards per day, and how many reviews?

Size new cards to your review hours, not to your motivation. The manual's own example: if you consistently learn 20 new cards a day, expect your daily reviews to settle around 200 cards a day. It warns that many users study hundreds of new cards in their first few days and then get overwhelmed by the reviews that follow.[1]

A simple way to turn that into your own number:

  1. Study for a week and work out your own average seconds per card (Anki's statistics screens show your time spent and cards reviewed).
  2. Multiply the manual's ratio: new cards per day times about 10 gives a rough steady-state review count.
  3. Divide your daily review time by your seconds per card to see how many reviews fit.

Worked example with made-up numbers (illustration only, not measurements): if you take 10 seconds per card and can give Anki 60 minutes a day, you can finish about 360 reviews, which supports roughly 36 new cards a day at the manual's 10-to-1 ratio. If you can only give 30 minutes, that falls to about 18 new cards a day.

For Maximum reviews/day, the manual says that once the limit is reached Anki stops showing review cards even if more are waiting, which smooths peaks and protects you when you come back after a week off.[1] Set it to the number you could finish on your worst study day. It is a safety rail, not a target.

Two related options are worth knowing. "Per-deck daily limits" let one preset serve many decks, with a Preset, This deck or Today only choice. And "New cards ignore review limit" is off by default; leave it off, because the manual says continuing to add new cards when you are behind makes a backlog worse.[1]

Which learning and relearning steps work with FSRS?

Keep them short and few. The manual says steps of one day or more are not recommended with FSRS, and that you should keep the number of steps to a minimum because repeating a card several times in one day does not contribute much to long-term memory. Examples that fit: 10m or 30m, or a pair such as 1m 10m.[1]

Why: the long steps many people used with the old scheduler were a workaround for "ease hell," where repeated failures lowered a card's ease. The manual says FSRS does not have that problem, and shorter steps let FSRS place the next review where it calculates the best time.[1]

What other settings are worth touching?

Only a few, and mostly for sanity:

  • Burying siblings. Bury new siblings, bury review siblings and bury interday learning siblings delay other cards from the same note until tomorrow. Useful for cloze-heavy decks where several cards come from one sentence.[1]
  • Display order. The manual says display order options come from the deck you select to study, not the deck of the card on screen. Its warning for backlogs is below.[1]
  • Easy Days. If you want lighter days (for example Sunday), Easy Days nudges due dates after the interval is calculated. It works with FSRS and SM-2, and it is not retroactive: you will not see an immediate change in due counts. Setting every day to Reduced or Minimum gives the same workload as Normal on every day.[1]
  • Maximum answer seconds. The internal timer caps at 60 seconds by default. The manual's advice if you often exceed it is to make cards simpler.[1]
  • Add-ons. The manual's rule of thumb: an add-on that changes intervals or scheduling should not be used with FSRS.[1]

What about the AnKing deck and AnkiHub settings?

This article does not quote AnKing or AnkiHub settings. We did not find an official AnKing or AnkiHub page stating a current recommended preset in our search on November 7, 2026, only community forum threads written by users, which disagree with each other. Check their own pages for current guidance, and avoid mixing presets from different sources. For deck choices see our guide to Anki decks for Step 1.

What are good starting points by phase?

These are our own starting points, not research-backed values and not Anki defaults. Use them as a first week, then adjust from your Stats. The only numbers taken from the manual are the 90% default, the "under 97%" caution and the sub-one-day steps.

Setting M1 and M2 (lecture-driven) Dedicated (Step 1 or Step 2 CK) Clerkships (M3)
FSRS On On, switched on weeks before dedicated, not the week it starts On
Desired retention 90% (the default) 90%; raise only if your simulator shows the workload fits 90%; lowering is a time trade, so check the simulator first
New cards/day Start near the manual's example of 20, raise only after two stable weeks Set from your hours after Qbank time; usually lower than in preclinical years A small number tied to patients you actually saw
Maximum reviews/day Your worst-day capacity Your worst-day capacity Your worst-day capacity on a long call day
Learning steps 10m, or 1m 10m Same Same
Easy Days Optional: a lighter weekend Usually off Reduced on your heaviest rotation days, if you know them in advance

Two practical points. First, shelf and Step questions create their own review load, so a new-card number that worked in M1 often has to drop once your Qbank time begins. Second, a rotation changes week to week; keep a second preset ready for call weeks and switch to it ahead of time, rather than letting reviews pile up and rescuing them later.

How do I dig out of a review backlog?

Stop adding new cards, cap your daily reviews, and work through what is due in an order that reduces forgetting. The manual's own advice: if you have a backlog of overdue reviews, stop introducing new cards until you catch up, and keep "New cards ignore review limit" off.[1]

A rescue plan using only manual-documented options:

  1. Pause new cards. Set New cards/day to 0 for the preset (or "Today only") while you catch up.[1]
  2. Set a daily ceiling. Set Maximum reviews/day to a number you can finish. Anki will not show more that day, even if more are due.[1]
  3. Change the review order. With FSRS on, the "relative overdueness" sort order is replaced by Ascending retrievability, which shows cards you are least likely to remember first. The manual describes relative overdueness as the generally recommended order for a large backlog under the older scheduler.[1]
  4. Do not restart from zero. Anki's Studying page says that when you answer a card that has been waiting a long time, it factors that delay into the next interval, so returning after a long break does not mean starting over.[3]
  5. Reintroduce new cards slowly once your due count is manageable. This step is our suggestion, not the manual's: restart at about half your old number and watch two weeks of Stats before raising it.

Do not respond to a backlog by switching on "Reschedule cards on change" or by changing retention drastically; both change due dates in ways that are hard to undo.[1]

Does Anki improve exam scores?

There is evidence that regular Anki use is associated with better Step 1 performance, but it is observational, and the picture for Step 2 CK is thin. A 2026 systematic review of 11 studies in Medical Science Educator found a consistent positive association between regular Anki use and USMLE Step 1 performance in three studies, with high-frequency users outperforming minimal users by 4 to 13 points. Only one study assessed Step 2 CK and found no significant benefit, and the authors note the evidence is largely observational.[4] Nothing in these studies says that any particular setting raises a score.

On the learning-science side, a randomised trial in Medical Education (Larsen, Butler and Roediger, 2009) found that repeated testing with feedback produced final scores 13 percentage points higher than repeated study when residents were tested more than six months after teaching (39% versus 26%).[5] That is the reason flashcard-style retrieval is worth the time; it is not a reason to pick any one number in Deck Options.

Where does Alma fit?

If you would rather not manage presets, Alma's Flashcards use FSRS scheduling out of the box. In Deck Options you can set new cards per day (20 by default), reviews per day (200 by default, with 0 meaning no limit), learning and relearning steps, and desired retention from 70% to 97% with a default of 90%. Alma can import an Anki .apkg file and keep the deck tree, and the Due Today and Stats views show your load.[6] Alma is a study tool, not a question bank of real exam questions, and it is not affiliated with Anki, AnKing, AnkiHub, the NBME or any exam body. See Alma Flashcards and the medical student page, or read how Alma's FSRS scheduling works.

Frequently asked questions

Should I use FSRS or the old scheduler?

The manual describes FSRS as an alternative to the legacy SM-2 scheduler that, by estimating more accurately what you are likely to forget, can help you remember more material in the same time.[1] It needs all your devices to support it, and it needs you to press Again (not Hard) when you forget. If those two are true, there is little reason to stay on SM-2.

What desired retention should I pick for dedicated study?

Start at 90%, the manual's default. Raise it only if the simulator shows a workload you can sustain; the manual says workload increases quickly above 90% and can overwhelm you above 97%.[1]

Why do I have so many reviews?

Usually because you added many new cards quickly. The manual's example is 20 new cards a day leading to roughly 200 reviews a day, and it names the "hundreds of new cards in the first few days" pattern as a common cause of overwhelm.[1] Lower new cards per day, set a maximum reviews per day, and see the backlog plan above.

Should my presets differ by deck?

Yes where the material differs. The manual says to give separate presets to decks that vary widely in difficulty, because FSRS parameters for easy decks differ from those for hard ones, and it notes parameters and retention are preset-specific.[1]

How is this different from a general Anki guide?

It is organised by what a medical student's year actually looks like. For a broader view of the tool, see our Anki guide for medical students, and for card quality see how to build flashcards that work. More medical student guides are in the medical hub.

Alma is a study tool. It does not provide medical advice and is not for clinical decisions. Never put patient information into any flashcard or study tool.

Sources

  1. Anki Manual, Deck Options (FSRS, desired retention, daily limits, steps, Easy Days, display order). https://docs.ankiweb.net/deck-options.html (read November 7, 2026)
  2. Anki, Download page (current desktop version 26.09.4). https://apps.ankiweb.net/ (read November 7, 2026)
  3. Anki Manual, Studying (the "Falling Behind" section). https://docs.ankiweb.net/studying.html (read November 7, 2026)
  4. Frappa N, Chernov D, Dillon M, Alben MG. Anki Use and Academic Performance in Medical Education: A Systematic Review of Evidence and Learning Theory. Med Sci Educ. 2026;36(2):1015-1025. PubMed PMID 42183420. https://doi.org/10.1007/s40670-026-02643-5
  5. Larsen DP, Butler AC, Roediger HL. Repeated testing improves long-term retention relative to repeated study: a randomised controlled trial. Med Educ. 2009;43(12):1174-1181. PubMed PMID 19930508. https://doi.org/10.1111/j.1365-2923.2009.03518.x
  6. Alma, Flashcards feature documentation (decks, FSRS, Deck Options, Anki import, Stats). Internal product facts, read November 7, 2026.