What Desired Retention Should You Choose in FSRS?

A desired retention of 85-90% is a reasonable default for most flashcard study. Go higher, toward 92-95%, for material you can't afford to forget, like an exam next month. Go lower, toward 80%, if you'd rather review less often and accept forgetting more cards along the way.

Short answer

• Desired retention is the recall probability FSRS targets at the moment each card comes due.
• Higher retention means shorter intervals and more total reviews.
• Lower retention means longer intervals, fewer reviews, and more forgetting.
• 85-90% is a reasonable default; adjust based on how costly forgetting a card actually is for you.
• Changing the setting reshapes future intervals; it doesn't retroactively rewrite cards you've already reviewed.

What desired retention actually controls

Desired retention is the recall probability FSRS aims for at the exact moment a card becomes due again. If you set it to 90%, the algorithm schedules each review so that, on average, you'd recall about 9 out of 10 cards correctly if tested right when they come due. For the mechanics behind how FSRS turns that target into an actual interval using stability and difficulty, see FSRS explained without the maths.

It's a target, not a guarantee for any individual card. Some cards will come due and you'll recall them easily; others will come due and you'll have forgotten them despite the target, because the underlying model is a prediction based on your history, not a certainty about any single review.

The trade-off: more reviews or more forgetting

There is no setting that gives you both fewer reviews and less forgetting — the two move in opposite directions, and desired retention is the dial between them.

Neither end of the range is "wrong." A higher target costs you time in exchange for fewer forgotten cards. A lower target saves time and accepts that you'll relearn some cards from a further-back starting point. The question isn't which number is objectively best — it's which trade-off matches what you're actually optimizing for.

How to pick a number for your situation

Match the setting to the actual cost of forgetting a card in your specific case, not to a general sense that "more retention must be better."

• An exam in a few weeks: lean higher, toward 90-95%. The review time cost is worth it because a forgotten fact on exam day is expensive in a way extra daily reviews aren't.
• A language you're maintaining over years: a moderate setting, around 85%, often makes more sense. You have time to relearn an occasionally forgotten word, and a lower review burden is more sustainable over years than months.
• A large deck you're worried about abandoning: consider a lower target, toward 80%, if the real risk to your retention isn't forgetting individual cards but quitting the habit entirely because reviews take too long every day. See how many new cards to add per day for the related question of pacing your review load.
• Material with irregular real-world stakes — professional certification content you'll need reliably on demand — generally warrants staying on the higher end, since the cost of a wrong answer in that context outweighs the extra review time.

Signs your retention target is wrong for you

• Reviews feel constant and exhausting, and you're skipping days because of it: your target is likely too high for the time you actually have. Lowering it reduces review frequency without abandoning the method.
• You're forgetting cards you feel you should know, and it's undermining your confidence before something important: your target may be too low for what's at stake. Raise it, especially as a deadline approaches.
• You changed the setting once and never checked whether it helped. Give a new target a couple of weeks before judging it — the effect shows up in your review count trend and your forgetting rate, not in the first session after the change.

Changing it later: what happens to existing cards

Changing your desired retention setting affects future scheduling decisions; it reshapes the intervals the algorithm calculates going forward, using the stability and difficulty already recorded for each card. It does not rewrite your past review history or pretend earlier reviews happened at the new target.

In practice, that means a change takes a little time to fully show up: cards you've already scheduled under the old target will gradually shift toward intervals consistent with the new one as they come due and get reviewed again, rather than jumping instantly. If you lower the target expecting an immediate drop in daily reviews, expect a gradual effect over the following weeks, not an overnight one.

How StudyDaily exposes this setting

StudyDaily runs FSRS-6 and lets you set your desired retention directly, rather than hiding it behind a fixed default you can't see or change. Each rating button shows the interval it will produce before you tap it, so raising or lowering your retention target has a visible effect you can watch immediately on individual cards, even while the overall review count shifts more gradually.

Reviews and scheduling run entirely on the device, so testing a new setting doesn't require a network connection or an account. Open StudyDaily in your browser to compare the interval a card gets at a higher versus lower retention setting — reviewing doesn't require signing up.

What should you do next?

Start at 85-90% if you're unsure, then adjust based on how the trade-off actually feels after a couple of weeks of real use.

• Set desired retention based on how costly forgetting a specific deck actually is, not a single universal number.
• Raise it temporarily as a deadline approaches; lower it if a large deck's review load is at risk of making you quit.
• Give a change two weeks before judging its effect — the shift is gradual, not instant.
• Watch your review count trend and your forgetting rate together, not either one alone.

Questions people ask

What is a good desired retention for beginners?

85-90% is a reasonable starting point: high enough that forgetting stays manageable, low enough that the daily review load stays sustainable while you're still building the habit.

Does a higher desired retention mean better learning?

Not automatically. It means fewer forgotten cards at the cost of more review time. If the extra reviews cause you to skip sessions or quit, a lower setting that you actually sustain produces better real-world retention than a higher one you abandon.

Will changing my desired retention affect cards I've already learned?

It changes future scheduling calculations using each card's existing stability and difficulty; it doesn't retroactively alter past reviews. The effect on your daily review count shows up gradually as cards come due and get rescheduled under the new target.

Should I raise my retention target before an exam?

Often yes, if the material is exam-critical and you have enough time before the exam for shorter intervals to fit your schedule. Raise it a few weeks out rather than the night before, since the benefit comes from more frequent reviews over time, not a last-minute setting change.

Sources used for this guide

The mechanics of how desired retention interacts with stability and difficulty come from the open-source FSRS documentation. StudyDaily's implementation claims come from the product's own FSRS-6 scheduler, checked against the source documentation for this guide's 20 July 2026 publication date.