How Alma uses FSRS to schedule your flashcards: Again, Hard, Good, Easy and what the intervals really mean

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

This article explains, in concrete terms, what each rating tells FSRS, what the shown intervals mean, and how to tune options like new cards per day, learning steps, desired retention, leech handling and study‑day rollover in a way that fits a real medical‑school week.^[1][2][3] We will also be explicit about what AI tools can and cannot safely replace in your studying.

What is FSRS and why does Alma use it for flashcards?

FSRS (Free Spaced Repetition Scheduler) is an open‑source algorithm that estimates how likely you are to recall each card and picks the next review date around a target retention rate, usually about 90%.^[4][5][6] It models each card with difficulty, stability and retrievability and schedules it for when your recall is predicted to drop to that target.^[4][5][7][6] Alma uses FSRS to schedule reviews but is not making any exam or memory guarantee.

FSRS is open and has been adopted by modern flashcard apps. Anki integrates FSRS as an alternative to its legacy SM‑2 algorithm, and users can enable it in deck options.^[1][2][6] The algorithm itself is maintained by the open‑spaced‑repetition community and is available for anyone to implement.^{[8][4][5][7][6]}

FSRS works by keeping per‑card parameters such as stability, difficulty and retrievability that describe the card’s memory state.^{[8][4][5][7][6]}

• Difficulty: how inherently hard the card is for you, on a scale from 1 to 10.^[5][7][6]
• Stability: a quantity that describes how slowly the memory fades and is represented in the FSRS model alongside difficulty and retrievability.^{[8][4][5][7][6]}

On each review, FSRS updates those values based on your rating and the time since the last review, then chooses the next interval so that your predicted recall when the card is due matches a target (the desired retention).^{[8][4][5][7][6]} Alma builds its spaced‑repetition features on this algorithm but, like other tools, cannot guarantee a given grade or pass outcome.

What do Again, Hard, Good and Easy actually mean in FSRS?

In FSRS, the four ratings describe two things: whether recall succeeded and how hard it felt.^{[9][4][10][5]} Again always means failed recall.^[9][4][10] Hard, Good and Easy all mean you recalled the answer, with Hard being a difficult success and Easy being effortless.^{[9][4][10][5]} FSRS uses that single reading to update difficulty and stability and pick the next interval.^[4][10][5]

• Again: you could not recall the required answer or got it wrong.^[9][4][10]
• Hard: you recalled it correctly after serious difficulty, doubt or hesitation.^[9][4][10]
• Good: you recalled it correctly with normal mental effort.^[9][4][10]
• Easy: you recalled it correctly with little or no meaningful effort.^[9][4][10]

FSRS and the tools that implement it stress that Again is the only rating that records a failure; the other three are gradations of success.^{[8][9][4][10][5][7][6]} Pressing Hard when you actually forgot tells the scheduler you succeeded, which can produce intervals that are unrealistically long for that card.^[1][9][4]

How do these ratings change the next interval?

Under FSRS, successful reviews use formulas that increase the card’s stability, while forgetting uses separate formulas for post‑lapse stability.^[8] Again collapses the interval and records a lapse, sending the card back to near‑term review in implementations that treat Again as a failed recall.^[2][9][10] Hard increases the interval by the least amount, Good by a standard amount, and Easy by the most, which can also make subsequent intervals grow faster.^{[8][2][4][10][5]}

The algorithm’s internals are expressed in formulas in the FSRS documentation and related explanations, but the practical picture is what matters when you are choosing ratings and settings.^[8][4][5]

• FSRS computes your current retrievability from stability and time since last review.^[8][4]
• It updates difficulty and stability based on your rating and how overdue the card was.^[8][4][5][7]
• It then finds the day when your retrievability will decay to your desired retention and sets that as the next due date.^{[8][4][5][7][6]}

Tools that surface FSRS often show the next interval for each button on the review screen. FSRS documentation emphasizes that those are previews of what the scheduler would do, not menu choices: you should pick the rating that reflects your recall, not the interval you like most.^[9][10]

Is it okay to use only Again and Good?

FSRS can work with all four buttons or with a simplified two‑rating habit. Both the Anki manual and FSRS tutorial note that using only Again and Good is valid: Again for every failed recall and Good for every success.^[1][9] This records less detail about difficult or effortless recalls, but the scheduler can still function.

That trade‑off can appeal if you find yourself overthinking Hard versus Good. The official FSRS guidance is that ratings should be based on whether you recalled the answer and how difficult that felt, not on specific time limits or displayed intervals.^[8][1][9][4]

What does “desired retention” mean and how should medical students set it?

Desired retention is the recall probability FSRS targets when it schedules a card’s next review.^[2][4][5][3] Higher desired retention means shorter intervals and more reviews; lower means fewer reviews and more forgetting.^{[1][2][4][5][3]}

Tools that expose FSRS options typically let you choose a desired retention within a limited range, and the documentation describes that higher values lead to more reviews per day.^[1][2][3] The Anki manual describes a default desired retention of 90% as offering a good balance of retention and workload, while warning that raising it above 97% can make the workload overwhelming.^[2]

For a medical student with dense content and fixed exam dates, a moderate desired retention offers a practical middle ground; pushing higher should be done cautiously because each small increase in target recall can produce a large increase in daily workload.^[2][4][5]

How do per‑deck options like new cards per day and learning steps interact with FSRS?

In apps that pair FSRS with traditional deck options, FSRS handles the memory model and long‑term scheduling, while deck options control aspects like card intake and early learning steps.^{[1][2][4][5][3]} Common settings include new cards per day, maximum reviews per day, learning steps and relearning steps.^[2][3]

Documentation for FSRS‑using tools illustrates some standard ranges:

• New cards per day: how many new cards can be introduced on a day you study that deck.^[2][3]
• Maximum reviews per day: an upper limit on review cards per day; further due cards are held for later.^[2][3]
• Learning steps: short, intra‑day delays like “1m 10m” that a new card cycles through before it graduates to long‑term review.^[2][3]
• Relearning steps: similar short steps used after a lapsed card gets Again.^[2][3]

The Anki manual notes that if you consistently learn around 20 new cards per day, you can expect daily reviews to reach around 200; if reviews pile up, reducing the new card limit is the safest first adjustment.^[2] Mnemo, another FSRS‑based tool, uses similar defaults and highlights that every new card generates future reviews, so intake should match your realistic capacity.^[3]

What about leeches and lapse handling?

FSRS distinguishes between successful and failed recalls in its formulas, and host apps can interpret repeated failures as “leeches”.^[8][1][2][9] Mnemo, for instance, lets you configure a lapse limit (number of Again ratings) before a card is tagged as a leech, with the option to automatically tag or suspend it when the limit is reached.^[3]

How do study‑day rollover, “Due today” and “Catch up” views work with FSRS?

FSRS operates over time, but host apps present reviews in daily batches and define when a “study day” starts, then group due cards accordingly.^[2][3] Some tools expose a configurable rollover hour for when the study day starts, with defaults such as 04:00 and a range from 00:00–23:00.^[3]

Mnemo explicitly lets users choose the clock time when a study day rolls over; its documentation gives 04:00 as the default, with valid settings from 00:00 to 23:00.^[3] This matters for late‑night study: a session after midnight but before the rollover can still count as part of the previous day, which keeps “Due today” counts more stable across irregular schedules.^[2][3]

How does Alma itself fit into a safe FSRS‑based study workflow?

Alma sits on top of FSRS as a study platform for medical, nursing, NP and PA students. It is a study tool, not an exam authority, and it makes no guarantees about memory or outcomes. Alma offers flashcards built from your own notes and uploads, all scheduled by FSRS, and combines them with quizzes, mind maps, slides and other tools drawn from your material.

FSRS itself is focused on scheduling; it does not decide what cards you write or whether those cards match your curriculum.^[4] Alma’s features are designed to help with that upstream work:

• Quizzes from your uploads: you can upload lecture documents and get questions written from them in the style of your exam; Alma reports what you miss so you can build flashcards for weak points.
• Notebook beside questions: while you answer, you can take notes that later become flashcards, diagrams or mind maps.
• Flashcards from your notes and uploads: Alma turns your own material into FSRS‑scheduled decks; each learner’s schedule is separate, even when decks are shared.
• Mind maps, handouts and Knowledge Graph: these help you organize complex systems before you compress them into card‑sized facts.

Because Alma uses your own sources, you still need to map them against primary references and official exam outlines. For guidance on safely mixing AI‑generated content with those primary sources, see our articles on turning lecture notes into good flashcards with AI and on using AI for USMLE Step 1 safely.

Alma is not a question bank of real exam questions and is not affiliated with the NBME, USMLE program, NBOME or any other examining body. It is not a clinical decision tool and is not medical advice. Instead, it automates the mechanical parts of building and scheduling cards from your own materials so that FSRS can handle the timing and you can focus on understanding.

Frequently asked questions

Are Alma’s intervals the same as Anki’s FSRS intervals?

FSRS is an open‑source algorithm that many tools can implement, including Anki and other flashcard apps.^{[8][1][4][5][7][6]} Each host app can still apply its own constraints, such as maximum intervals, daily review caps or learning steps.^[2][3] Alma uses FSRS for scheduling but is a different product from Anki with its own interface and defaults.

What happens if I often hit Hard instead of Again when I truly forget?

The FSRS tutorial explicitly warns that pressing Hard when you have actually forgotten will cause the scheduler to treat that review as a success and may make future intervals unreasonably long.^[1][9][4] To keep the schedule realistic, choose Again whenever you cannot recall the required answer or get it wrong.^[9][4][10]

Can I change desired retention per deck for different kinds of material?

FSRS supports choosing a desired retention, and tools that expose FSRS settings usually let you adjust it at the level of a preset or deck.^[1][2][3] A higher target brings more frequent reviews and fewer lapses; a lower target stretches intervals at the cost of more forgetting.^{[1][2][4][5][7][3]}

What should I do when my review backlog gets out of control?

The Anki manual recommends stopping the introduction of new cards until you clear a backlog of overdue reviews, because each new card creates future review load.^[2] FSRS can handle overdue cards mathematically by accounting for the longer elapsed time, but human workload is still finite.^[8][4][5][7] Using features like daily review caps, backlog views or catch‑up modes can help stabilize your routine.^[2][3]

References

1. fsrs4anki/docs/tutorial.md at main · open-spaced-repetition/fsrs4anki · GitHub
2. Deck Options - Anki Docs
3. Review settings | Docs | Mnemo
4. What Is FSRS? How the Spaced Repetition Scheduler Works (2026) | Nibomo
5. FSRS Explained: The Algorithm Behind Smarter Flashcards — flashcards.sh
6. FSRS Algorithm Explained: Smarter Flashcard Scheduling | Lexkite
7. Spaced repetition: seeing each question again at the right moment · Erudiam
8. The Algorithm · open-spaced-repetition/awesome-fsrs Wiki · GitHub
9. Again vs Hard in FSRS (2026): Use the Four Flashcard Ratings | Nibomo
10. Again, Hard, Good, Easy: what your rating actually changes — Kikusho