You opened your deck this morning and something felt off. Cards you reviewed yesterday are suddenly due next month, your daily count dropped to almost nothing, or a pile you had spaced out neatly got shoved deep into the future. Your app did not ask first. It just rescheduled. In 2026 this happens all the time, usually after an update, a switch to a smarter scheduling algorithm, or a sync that landed differently on two devices. The good news is that your review history is almost certainly intact, and the schedule you are staring at is fixable. The catch is that the most tempting fix, a full reset, is the one that can actually cost you progress. Here is how to figure out what changed, protect what you have, and get the schedule back under your control without wiping months of work.

Don’t Panic-Reset: an app that started auto-rescheduling has almost never deleted your progress. Your review history and card maturity are stored separately from the due dates you see, so the schedule can move while the memory data stays put. Before you change anything, make a backup. Then find out what changed: a newly enabled scheduling algorithm like FSRS, a fresh optimization of its parameters, an app update with new defaults, or a sync conflict between devices. Fix the cause, roll back from a backup if the shift looks wrong, and leave the “Reschedule cards on change” option off unless you have a specific reason to flip it.

The first thing to understand is that due dates are a calculation, not something you set by hand. When the inputs to that calculation change, the outputs move. Most modern flashcard apps in 2026 run a predictive scheduling algorithm, and in Anki that algorithm is FSRS, which estimates when you are likely to forget each card and sets the next review accordingly. Turning that algorithm on for the first time, re-optimizing its parameters against your review log, adjusting how confident you want to be that you will remember a card, or installing an update that ships new defaults all change the inputs. The app then recomputes due dates to match. That recompute is the “auto-rescheduling” you noticed. In most cases it is not random and it is not a bug. It is the scheduler doing fresh math on new settings. If you are still comparing which app to keep your cards in, our roundup of the best flashcard apps of 2026 covers how each one handles this kind of automatic scheduling.

Not every reschedule has the same cause, and the exact pattern points you to the fix. Before you edit a single setting, look at what actually shifted and match it to the table below. This keeps you from flipping five switches and never learning which one mattered.

The single most important move is a backup, because every fix below is safer when you can return to the exact state you have right now. Anki keeps automatic backups and also lets you make a manual one on demand, as its own backups documentation explains. Make a manual backup now, and note the timestamp of the most recent automatic backup from before the reschedule happened, since that earlier snapshot is your rollback point. On a phone app, export your collection or confirm your cloud sync has a restore option before you start experimenting. A backup costs you about thirty seconds and turns a change that feels irreversible into one you can walk straight back.

Open your deck’s scheduling settings and read them before you edit anything. In Anki that lives in Deck Options, and three settings drive almost every surprise reschedule. The first is whether FSRS is switched on at all. The second is your desired retention, which is the probability you want of remembering a card at the moment it comes due. The common default sits around 0.90, and lowering it stretches your intervals so more cards get pushed out, while raising it pulls them back in. The third is the “Reschedule cards on change” checkbox, which is off by default for a reason we will get to. Anki’s deck options guide walks through each field. Change one setting, watch what it does to your due counts over a day, then decide. Editing three at once is exactly how people end up unable to tell which knob caused the mess.

This is the setting people misunderstand most, so it is worth slowing down on. Leaving it off does not mean your new algorithm is being ignored. It means future reviews use the new scheduling while the cards already sitting in your queue keep their current due dates until the next time you answer them. Turning it on tells the app to recompute the due date of every card immediately, all at once, which is what produces that dramatic overnight shift. Flipping it on carries two costs. The obvious one is the sudden pile of cards that all move together. The quieter one is that a full reschedule writes a new review entry for every card, which inflates your database and, on some setups, your sync payload. Because a normal optimization already improves future scheduling without any of that, the default-off behavior is the calmer path for most people. This kind of automatic behavior is showing up across the wider crop of AI study tools in 2026, so it helps to know what the toggle really controls before you use it. Switch it on only when you deliberately want every existing card re-timed in one pass and you already have a backup ready.

If the new schedule looks wrong and you would rather go back to yesterday, you have two clean options. The first is to restore the automatic backup from just before the change, which puts your due dates and everything else back to that exact point. The tradeoff is that you also lose any reviews you did after that snapshot, so choose the most recent pre-change backup you have. The second option applies when the trigger was an app update that shipped new behavior: revert to the previous app version, let it re-sync, and then re-enable features on your own timing instead of all at once. For a reschedule you mostly like but want smoothed out, a community add-on such as the FSRS Helper can reschedule and disperse due dates without expanding your database the way the built-in mass reschedule can, which spreads the review load across days instead of dumping it on one.

A common 2026 version of this problem is a schedule that looks fine on your laptop and wrong on your phone, or the reverse. That is a sync conflict, and the risky moment is the prompt that asks which side to keep. Choosing to download or upload the wrong direction can overwrite your good collection with a stale one, and that is the scenario that actually loses progress. Before you answer that prompt, know which device holds your most recent reviews, and back up that device first. If you are not sure, cancel the sync, back up both sides, and resolve it deliberately instead of clicking through under pressure. The rule that keeps you safe is short: never force a one-way sync that overwrites the device with your newest history until that device has a backup. Get the direction right and your progress survives the conflict completely untouched.

Here is the part that saves you the most stress. A modern scheduling algorithm is built to be self-correcting. If it schedules a card too far out and you forget it, that lapse feeds back into the model and tightens the next interval. If it is too aggressive, the same feedback loosens things up. Re-optimizing your parameters roughly once a month, after you have a few hundred reviews logged, keeps it tuned to how you actually remember rather than to a generic starting profile. So once you have backed up, audited your settings, and confirmed nothing destructive happened, the smart move is often to leave it alone for two weeks and let the schedule prove itself. Most reschedules that feel alarming on day one look reasonable by day ten. If your real problem is a backlog you created by missing days rather than an app-driven shift, our guides to rebuilding spaced repetition after a missed week and the spaced-repetition catch-up calculator will get your daily counts back to sane, and our list of the best study planner apps of 2026 can help you spread the load around the rest of your week.

Almost certainly not. In apps like Anki, your review log and each card’s memory data are stored separately from the due dates you see on screen. A reschedule recalculates when cards appear next, but it does not erase how many times you have answered them or how mature they are. If maturity itself looks reset, that points to a full reset or a bad sync rather than a reschedule, which is why a backup before you experiment matters so much.

Usually no. Leaving it off still applies your new algorithm to future reviews, so you get the benefit without a mass shift. Turning it on recomputes every card’s due date at once, which floods your queue and writes a new review entry per card that bloats your database. Enable it only when you specifically want every existing card re-timed in a single pass, and only after you have made a backup.

That is the classic signature of FSRS being enabled or freshly optimized, often paired with a lower desired retention. A lower retention target tells the scheduler you are comfortable reviewing less often, so it stretches intervals and pushes cards out. Open your deck options, check whether FSRS was just switched on, and raise desired retention closer to 0.90 if you want cards to return sooner.

Restore the automatic backup from just before the change, which returns your due dates to that exact point. You will lose any reviews done after that snapshot, so pick the most recent pre-change backup. If an app update caused it, revert to the previous version and re-sync. To keep the new scheduling but smooth the pile out, an add-on like FSRS Helper can disperse due dates without inflating your database.

No. Optimizing your parameters improves how future reviews are scheduled on its own, with no mass reschedule required. The optimization tunes the model to your history, and each card picks up the improved timing the next time you answer it. Running a full reschedule on top of that is optional and mostly useful when you deliberately want every current card re-timed immediately rather than gradually.

• Make a backup before you change anything, and note the last automatic backup from before the reschedule as your rollback point.
• Match the symptom to the cause: mass push-out means FSRS or retention, a same-day pile means “Reschedule cards on change”, device disagreement means a sync conflict.
• Audit deck options and change one setting at a time so you can see exactly what each does to your due counts.
• Leave “Reschedule cards on change” off unless you truly want every card re-timed at once, and have a backup ready if you flip it.
• Resolve sync conflicts deliberately, backing up the device with your newest reviews before you ever force a one-way sync.

An app that reschedules your cards without asking feels like it took the wheel, but it almost always left your progress exactly where it was. The due dates moved because the math behind them moved, and both the cause and the cure are within reach. Back up first so every choice is reversible, read your deck options before you edit them, and change one thing at a time. If the shift was wrong, roll back cleanly from a snapshot; if it was mostly right, give it two weeks to prove itself before you touch it again. A predictive scheduler earns your trust by correcting itself as you review, and the whole point of comparing AI flashcards against manual ones is deciding how much of that automation you want in the first place. Keep the control, keep the history, and let the schedule do the work it was designed to do.