How Many New Anki Cards Per Day: Computing Your Sustainable Ceiling
How many new Anki cards per day you can sustain is not a fixed number to copy from a default. It is a figure you compute from your own time budget. Set it too high, and the review load that arrives weeks later will outrun the time you are willing to give. That is the single most common reason learners abandon a deck.1
Overview
The optimal new-card load in Anki is the daily intake rate whose later review cost still fits inside the minutes you will genuinely spend reviewing. The sustainable Anki daily new-card limit is personal. It depends on your budget, your retention setting, and the kind of cards you study, not on a universal best number.
Why this is the question that breaks most decks
The new-cards/day setting controls how many new cards Anki introduces each day you open the program.1 Studying new cards "will temporarily increase the number of reviews you need to do a day," because "newly-learned material needs to be repeated a number of times before the delay between repetitions can increase appreciably."1
That deferred cost is the trap. Week-one load feels light, so a high intake rate looks sustainable right up until the reviews from those cards come due.
The manual's own fix is to cut intake, not push through: you "decrease the reviews required by introducing fewer new cards each day until your review burden decreases."1 The lever for review load is the new-card rate, and that single fact is what this article turns on.
The right new-card rate is computed from a personal review budget, not copied from a default. A default that works for one learner can wreck another's schedule three to six weeks in, because their budgets differ.
The standard range, and why a range is the honest answer
The commonly cited starting range for language learners is roughly 10 to 25 new cards per day, occasionally up to 30. One language-learning write-up gives 20 to 25 as the standard range, noting that "20 is manageable for most people, 25 may be too much for some."2
Treat that range as a community and practitioner heuristic, not a measured optimum. The Anki software default is 20 new cards per day, but the manual does not claim 20 is optimal for any learner. It only documents what 20 costs.13
The range is a starting bracket, not an answer. The rest of this article gives the arithmetic that turns the bracket into your number.
The math: why a new card is a loan, not a purchase
A new card costs almost nothing today
On the day a card is introduced, it sits in learning steps and gets repeated a few times in that one session. The cost of adding one card today is small, because "newly-learned material needs to be repeated a number of times before the delay between repetitions can increase appreciably."1
The growth in workload is delayed, not absent. Week-one load reflects only cards added in week one. It does not yet include the recurring reviews those cards will generate over the following weeks.
At steady state it bills you every month
Here is the durable, officially sourced figure this article rests on. The Anki manual states: "If you are consistently learning 20 new cards a day, you can expect your daily reviews to be roughly about 200 cards/day."1
That is roughly a tenfold multiplier from new cards per day to steady-state daily reviews, stated by the official manual.1 Adding cards is like taking out a loan whose monthly payment comes due later.
Learners often explain the same effect with a rule of thumb: each new card generates about seven reviews over the following month, so 20 new per day climbs toward roughly 140 daily reviews within a month. Treat both the "about 7 reviews per card" and "about 5 to 8 reviews per month at steady state" figures as community heuristics, not constants.1
The Anki manual's 20-new to roughly-200-review relationship is the citable anchor. The per-card monthly numbers are the informal intuition learners use to explain it.1 Either way, each mature new card settles into a recurring review tax, and 20 new per day compounds into a triple-digit daily review count within weeks.
The 3 to 6 week lag is why people get blindsided
The load from a batch of new cards does not land the day you add them. It builds over the following weeks as those cards cycle through learning and early review intervals before their intervals stretch out.1
Reviews from cards added across several past weeks all come due in overlapping waves. Daily load therefore keeps climbing for roughly the first month or more before it plateaus. That plateau is the steady-state figure, not the comfortable week-one number.
The pain arrives after the honeymoon, precisely when motivation has normalized into routine. The three-to-six-week window is a practitioner observation about when steady state is reached, not a measured constant; treat it as a lag estimate.
A picture makes the delay concrete. The following diagram traces how daily review load grows week by week from a constant new-card intake, then levels off once the earliest cards mature out.
Computing your sustainable ceiling
Start from your weekly time budget, not a card count
Because daily reviews scale with new cards per day, you can run the manual's relationship backward.1 Decide how many minutes per day you will genuinely give reviews. A 7-day time audit shows what that figure realistically is. Then convert that time to a daily review ceiling, and back out the new-card rate that lands under it.
You need a per-card time estimate to do the conversion. A fast single-word recognition card often clears in under ten seconds. A sentence or production card can take two to three times that, so the only reliable figure is the one you measure on your own deck.
Treat any per-card number as illustrative, not a constant. Time your own deck for one week before trusting any figure, because heavier cards run slower.
The manual documents a "Maximum reviews/day" option that sets "an upper limit on the number of review cards to show each day. When this limit is reached, Anki will not show any more review cards for the day, even if there are more waiting."1 A new-card limit plus a review cap are the two built-in guards that hold your computed ceiling in place.
The variables that move your number
Your desired-retention setting drives reviews per card, and that relationship depends on which scheduler your deck runs. SM-2 vs. FSRS: How Anki's Scheduling Algorithms Work traces why. The Anki manual states: "Higher retention leads to shorter intervals and more reviews per day. The default is 90%, which offers a good balance of retention and workload. Above 90% the workload increases very quickly, and above 97% the workload can be overwhelming."1
The FSRS project frames the same trade-off from both directions. As desired retention increases, "intervals become shorter, which increases the number of reviews per day." As it decreases, "we forget more and must re-learn more of our material."4
There is therefore a point of minimum workload that balances the two. The FSRS project notes that this optimum varies by learner and does not publish a single universal percentage.4
Card complexity is the second variable. Production cards and sentence cards take longer per review and tend to lapse more than single-word recognition cards, which raises effective reviews per card.
Outside reinforcement is the third. Real Japanese reading and listening alongside the deck reinforces the same items, lowering the effective failure rate and therefore the review load per card.4 This effect has a direction, but not a fixed size. Do not expect a fixed percentage.
A worked example
The arithmetic below uses the manual's anchor relationship as its spine, and every number is illustrative. The method is the point, not the placeholders.
Suppose you decide to give reviews 20 minutes per day. At roughly 10 seconds per review card on a light recognition deck, that budget buys about 120 review cards per day as your steady-state ceiling.
Now back out the new-card rate. The manual's 20-new to roughly-200-review relationship implies about a tenfold multiplier, so a 120-review ceiling supports roughly 12 new cards per day.1
The card type changes the result sharply. If your cards run about 20 seconds each, the same 20 minutes buys only about 60 reviews per day, which implies roughly 6 new cards per day.
| Per-card review time | Reviews from a 20-min budget | Implied new cards/day |
|---|---|---|
| ~10 sec (light recognition) | ~120/day | ~12/day |
| ~20 sec (heavier/sentence) | ~60/day | ~6/day |
Same budget, half the new-card rate, because the per-card cost doubled. Measure your own per-card seconds and your own steady-state multiplier, which rises with desired retention. Do not rely on these placeholders.14
The failure mode and the diagnostic
What collapse looks like
When intake outruns budget, the review queue grows because new cards keep adding delayed load faster than old cards mature out. The manual's prescribed fix assumes exactly this failure: "decrease the reviews required by introducing fewer new cards each day until your review burden decreases."1
The lived pattern is plainer than the mechanism. Load outruns the willing budget, sessions stop being rewarding, a backlog grows, and eventually the learner abandons the deck, not just one session.
That last step is the cost worth naming. The deck dies not because the method failed, but because the intake rate was never affordable.
The two-week diagnostic
The test is to watch whether your total daily review time exceeds your willing budget for a sustained stretch, roughly two weeks. If it does, your new-card rate is probably too high.
The two-week window is a practitioner heuristic for "long enough that this is a trend, not a single bad day," not a published protocol. Treat it as a rule of thumb, not a sourced constant. If the sessions themselves have already become painful rather than merely long, Why Anki Has Become Painful (and How to Fix It): A Diagnostic Triage runs the broader triage.
The manual's prescribed response is to lower the new-card limit, not to power through: introduce "fewer new cards each day until your review burden decreases."1
How to recover without nuking progress
Lower the new-card limit toward zero temporarily so no new load is added. Let the existing backlog drain as those cards mature and their intervals lengthen. Then resume at a rate the budget actually supports.1
The lever is the same new-cards/day setting the manual identifies, and setting it to zero is a valid configuration rather than an error state.1 For deeper recovery involving leeches and card suspension, When SRS Becomes Counterproductive: Anki Burnout, Leeches, and the Exit Signs covers the protocol.
Good to know
"Zero new cards for a while" is a valid setting, not a failure
The new-cards/day setting accepts any value down to zero. Setting it to zero simply introduces no new cards while reviews continue to drain.1 Pausing intake to clear a backlog is normal deck hygiene, not a sign the method failed.
The manual frames the new-card rate as the dial you turn down to reduce review burden. That logically includes turning it all the way to zero while a backlog clears.1 Resume intake once the queue is back inside your budget.
Sentence-mining decks change the arithmetic
A common error is counting one self-mined sentence card as equivalent to one pre-made vocabulary card. They are not workload-equivalent. Longer, denser cards run slower per review and tend to lapse more. Higher reviews per day follow from material that must be "repeated a number of times."1
FSRS likewise ties review count to per-card difficulty and retention.4 The practical consequence is that the sustainable new-card ceiling for a sentence-mining deck should be lower than for a pre-made vocabulary deck. This is a directional adjustment. There is no fixed multiplier to apply.
A hard daily cap protects you from the honeymoon
A fixed new-card limit plus a parent-deck review cap stop week-one enthusiasm from writing a check week six has to cash. The manual documents both guards: the new-cards/day limit, and a "Maximum reviews/day" cap that stops showing review cards once the limit is hit.1
For nested decks, "the limits set on each subdeck control the maximum number of cards drawn from that particular deck. The selected deck's limits control the total number of cards that will be shown." In other words, a parent-deck cap bounds the whole study session.1
One interaction is worth knowing. "By default, the review limit also applies to new cards, and no new cards will be shown when the review limit has been reached," unless you enable the option to show new cards regardless.1 A single review cap can therefore quietly slow new intake too.
The arithmetic above assumes you have already decided what to study. Building and leveling a deck is its own drain on the same time budget. For learners who would rather spend their minutes reviewing than curating, J-Compass recommends Amenokori: its FSRS-scheduled, pre-leveled catalogue removes the deck-building work, so the main dial left to manage is the daily new-card rate this article teaches you to set.5 Its published per-level entry counts (for example N5 801, N3 3,355, N1 3,239 plus 803 extended) are Amenokori's own leveling of its catalogue, not official JLPT figures, since no official JLPT vocabulary list has been published since 2010.5
Beware the 50 to 100 per day numbers from full-immersion learners
New-card rates of 50 to 100 per day circulate as personal achievements from learners in their first months of intensive study, including one who sustained 100 new per day for a full month.2 Those rates come from contexts where the deck is the main daily activity with no competing study load, so copying them into a part-time schedule is a mistake.
By the manual's own scaling, 50 new per day projects toward roughly 500 reviews per day at steady state, extrapolating the 20-new to roughly-200-review relationship.1 That is a multi-hour daily commitment, incompatible with a working adult studying Japanese a few hours a week. The high numbers are not a template. They assume a budget most learners do not have.
See also
- Spaced Repetition and the Forgetting Curve: Why Reviewing on a Schedule Works
- Beyond Anki: SRS Tools and Approaches Compared
- Hours per Day vs. the Marathon: Pacing Your Japanese Study
- A Daily Kanji Study Routine: How Many Kanji per Day, Review-Load Math, and the Three-Block Schedule
- Yomitan + Anki: One-Click Card Creation