How does FSRS decide when to schedule the next review?

You set a desired retention (typically 0.90); FSRS predicts how R decays and schedules the review for the day R falls to that target.

Because stability S is defined as the days for recall to drop to 90%, aiming for 0.90 retention roughly means "review each card about one stability-length after the last review." Raise your target retention → reviews come sooner and more often; lower it → longer gaps, fewer reviews, more forgetting tolerated.

FSRS then fits its parameters to your own review history with machine learning, so the schedule is personalised to how you forget rather than one-size-fits-all.