Flashcards have a misleading reputation: people think they are only for foreign vocabulary. In fact they suit almost anything you need to recall, from anatomy to constitutional law. But a badly written card is useless, or worse, it gives you the illusion that you know something. Here is how to write cards that stick, with concrete examples. If you would rather start from a ready-made set, Mindlet generates flashcards from your course material, which you can then check against the same rules.
Why do flashcards work?
A flashcard makes you retrieve an answer before you see it. That act, active recall, is one of the two best-supported learning techniques, along with spaced repetition (Dunlosky et al., 2013, Psychological Science in the Public Interest 14(1), 4-58). Rereading a summary feels like knowing; testing yourself shows what you actually know.
Cards have a second advantage: you can shuffle them and spread them out over time. Nate Kornell compared two ways of using the same vocabulary cards: one big stack, where each card comes back after many others, and several small stacks studied one after another. The big stack, in other words spacing, won for 90 percent of participants (Kornell, 2009, Applied Cognitive Psychology 23(9), 1297-1317).
What makes a good card?
Piotr Woźniak, the Polish researcher who created the SuperMemo spaced repetition software, wrote twenty rules for formulating cards in 1999 (Twenty rules of formulating knowledge). The most useful ones come down to six principles.
- Understand before you memorize. A card locks in what you have understood. It does not replace understanding. If you cannot explain an idea, go back to the material before you make a card.
- One idea per card. Woźniak calls this the minimum information principle: the simpler the question, the more stable the memory. A card that mixes three ideas will always be half known.
- A precise question. There should be one expected answer. “Tell me about the French Revolution” is not a flashcard question.
- No lists. A seven-item list is one of the hardest things to remember from a card. Split it, or turn it into a cloze deletion where one item is missing at a time.
- Your own words. Rephrasing forces you to process the information. Psychologists call this the generation effect: we remember what we produce better than what we only read (Slamecka and Graf, 1978, Journal of Experimental Psychology: Human Learning and Memory 4(6), 592-604).
- Context when you need it. An example, an image or a chapter cue stops two similar cards from blurring together.
What does a bad card look like, and how do you fix it?
Rules make more sense with examples. Here are five cards as people often write them, and a better version of each.
| Bad card | The problem | Better card |
|---|---|---|
| Q: Photosynthesis? A: A whole paragraph | Vague question, answer too long | Q: What are the two products of photosynthesis? A: Glucose and oxygen |
| Q: Causes of World War I? A: Seven causes | Long list | One card per cause, or a cloze: “Archduke Franz Ferdinand was assassinated in … in June 1914.” A: Sarajevo |
| Q: Judicial review? A: A definition pasted from a website | Copied, not rephrased | Q: Which 1803 US Supreme Court case established judicial review? A: Marbury v. Madison |
| Q: True or false: mitosis produces two cells? | A coin flip gets it right half the time | Q: How many daughter cells does mitosis produce, and with how many chromosomes? A: Two, with the same number as the parent cell |
| Q: Derivative of ln(x)? A: 1/x | Correct, but isolated | Keep it, and add a card that applies it: “Derivative of ln(3x)?” A: 1/x |
Notice the last row: a correct card can still be too thin. In math or physics, add application cards, and practice full problems as well.
How should you review your flashcards?
A good card reviewed badly does not help much either. Four habits make the difference.
- Answer before you flip. Out loud, on paper or in your head, but a real answer. “I think I know this” does not count.
- Grade yourself honestly. If the answer came with hesitation, you do not know it yet. In Mindlet, three buttons ask you after each flashcard (in French, since the app interface is in French): « Acquis » (got it), « Flou » (fuzzy) or « Oubli » (forgot). “Fuzzy” is the most useful answer: it brings the card back sooner without treating it as forgotten.
- Do not drop the cards you know. This is the least intuitive lesson. In Karpicke and Roediger (2008), Science 319(5865), 966-968, students learned 40 Swahili words. Those who kept being tested on words they had already recalled remembered about 80 percent a week later. Those who dropped them from testing remembered only 33 to 36 percent, even when they kept restudying them.
- Space your reviews. A card you get right comes back later; a card you miss comes back soon. That is spaced repetition, and an app can do the scheduling.
Should you write your own cards?
Writing your own cards is excellent practice, for the reason above: rephrasing is already learning. But it is slow. For a semester of medicine or law, writing every card by hand can take longer than reviewing them.
A compromise that works: let AI build a first set from your material, then work on it. Read the cards, rewrite the ones that do not make sense to you, add the ones that are missing. You keep much of the benefit of rephrasing without spending your evenings copying.
In Mindlet, you add a PDF, a slide deck, a video or a link, and the AI prepares an anchor card for each key idea (a flashcard or a multiple-choice question), then formats suited to the subject: cloze deletions, ordering, matching, open questions and more. The PDF to flashcards page shows how it works. While the app interface is in French, your cards can be generated in nine languages, English included. You can also write your own cards one by one, and existing Anki decks can be imported (.apkg and .colpkg files): see Anki import.
The most common mistakes
- Making cards for everything, including what you already know perfectly.
- Copying the textbook word for word on the answer side.
- Creating a hundred cards the night before the exam, with no time to space them.
- Always reviewing in the same order, which teaches you the sequence rather than the answers.
- Marking a card as known because you are tired of it.
What makes flashcards work is the effort of retrieving the answer, repeated at the right moment. To go further on that mechanism, read our article on active recall and the testing effect.
FAQ
Frequently asked questions
How do you make a good flashcard?
Ask one precise question with a short, unambiguous answer. Avoid lists: split them into several cards or use a cloze deletion. Write in your own words rather than copying the textbook, and add an example or some context when the idea is abstract.
How many flashcards should I make per chapter?
There is no ideal number. Make one card for each thing you need to be able to recall: a definition, a date, a formula, a cause and its effect. Twenty precise cards beat five overloaded ones, which are hard to review and give a false sense of mastery.
Should I stop reviewing cards I already know?
No. In Karpicke and Roediger (2008), students who kept testing themselves on words they had already recalled remembered about 80 percent a week later, against 33 to 36 percent for those who dropped them from testing. Known cards should come back less often, not never.
Is it better to make flashcards by hand or with AI?
Writing your own cards helps you learn, but it takes time. AI can produce a first set from your course material in minutes, which you then review and complete. What matters most is answering from memory before you flip the card, and reviewing at growing intervals.
