How to Make Flashcards That Actually Work
TL;DR
Good flashcards ask for one thing, in your own words, in a way you can't answer by recognition. Split anything with an 'and' in the answer into separate cards. Write the question so there's exactly one correct response, and write the answer as the shortest thing you'd accept if someone said it out loud. If a card keeps failing, it's almost always too big or too vague — rewrite it rather than grinding it.
A flashcard is a promise you make to your future self: you will be able to produce this. Most cards break that promise, and they break it in predictable ways. The fix isn't discipline or more reviews — it's writing better cards in the first place. Here is what separates a card that holds for years from one you'll delete in frustration.
1. One idea per card
This is the rule that fixes most bad cards on its own. If the answer contains an and, a list, or a comma-separated series, it's more than one card pretending to be one. A card with three facts on it fails if you miss any one of them, so it will feel permanently hard — and worse, the two parts you do know get dragged back for review alongside the one you don't.
| Too big | Split into |
|---|---|
| What are the three branches of the US government and what does each do? | Three cards: one per branch, each asking what that branch does. |
| What does the Krebs cycle produce? | One card per output, or a cloze on the reaction you actually need. |
| Define 'photosynthesis' (full textbook definition) | One card for the inputs, one for the outputs, one for where it happens. |
The test
Read your answer out loud. If you need a breath in the middle, it's two cards.
2. Make the question impossible to answer by recognition
The whole value of a flashcard comes from retrieval — dragging the answer out of memory. A question that can be satisfied by a vague sense of familiarity gives you the feeling of knowing without the strengthening. This is the difference between active recall and simply looking at your notes again.
- "Is the mitochondrion the powerhouse of the cell?" — a yes/no card you'll get right by guessing. Useless.
- "What does the mitochondrion do?" — forces production. Better.
- "Which organelle produces most of the cell's ATP?" — forces production and is specific about which answer counts. Best.
3. Write the question so only one answer is right
Ambiguous questions are the quiet killer. "Tell me about the French Revolution" has no correct answer, so every review is a judgement call and you end up rating yourself on mood. If you find yourself thinking well, sort of when you flip a card, the question is underspecified. Add the constraint that makes it answerable: a year, a person, a mechanism, a single number.
4. Use your own words
Copying a definition verbatim from a textbook feels rigorous and produces cards you can recite without understanding. Worse, textbook phrasing is often dense on purpose, so you end up memorising syntax instead of meaning. Rewrite the answer as you'd explain it to a friend who asked. If you can't, you haven't understood it yet — and that's worth knowing before you spend six months reviewing it.
“Conditions that create difficulties for the learner can enhance long-term retention.”
5. Give the card enough context to stand alone
You'll meet this card in six months, out of order, with no memory of the chapter it came from. "What was the second cause?" means nothing then. Name the subject in the question itself — "In the 1929 crash, what was the second major cause of bank failures?" — even though it feels redundant while the material is fresh. It will not feel redundant later.
6. Use cloze deletions for anything embedded in a sentence
Some facts only make sense in context: a line of a formula, a word in a foreign phrase, a step inside a sequence. For those, blank out the single element you want to recall and leave the rest of the sentence visible. It keeps the context that makes the fact meaningful while still demanding one specific retrieval. Just resist blanking three things in the same sentence — that's rule one again, wearing a disguise.
7. When a card keeps failing, rewrite it — don't grind it
Every deck develops a few cards you get wrong over and over. The instinct is to review them harder. Almost always, the card is the problem: it's testing two things, or the question is ambiguous, or the answer is phrased in words you never actually use. Rewriting a leech takes thirty seconds and saves months of failed reviews.
A useful threshold
If you've forgotten the same card four or five times, stop rating it and edit it. Split it, sharpen the question, or add the context that makes the answer obvious once you know it.
How many cards should you make?
Fewer than you think. Every card is a small recurring commitment — a few seconds a day, forever, until you delete it. A chapter rarely needs fifty cards; it usually needs the eight or ten facts you'd be embarrassed to forget. Be ruthless about what earns a card, because your daily review queue is the bill for that decision.
This is also where writing cards by hand becomes the bottleneck. knojo drafts cards from your notes or an article — one idea per card, phrased as questions — so the work shifts from typing to editing. You keep the good ones, fix the vague ones, and bin the rest. Curating is much faster than writing from nothing, and the rules above are exactly what you're curating for.
Question
What does the spacing effect say about memory?
Answer
We remember more when reviews are spread out over time than when they're crammed together.
Fix five cards today
Open your deck and find the five you get wrong most often. Don't review them — rewrite them. Split anything with an 'and', sharpen anything ambiguous, and put the subject back into the question. That's usually the whole problem.
Frequently asked questions
What makes a good flashcard?+
A good flashcard tests one idea, asks a question with exactly one correct answer, uses your own wording rather than copied text, and includes enough context to make sense months later out of order. If the answer needs a breath in the middle when read aloud, it should be split into separate cards.
How many facts should be on one flashcard?+
One. This is sometimes called the minimum information principle. A card with several facts fails whenever you miss any one of them, so it stays permanently difficult and drags the parts you already know back into review alongside the part you don't.
Should I copy definitions from my textbook onto cards?+
No. Verbatim definitions let you recite wording without understanding the idea, and textbook phrasing is often deliberately dense. Rewrite the answer the way you would explain it to a friend. If you can't rephrase it, you don't understand it yet — which is useful to discover early.
What is a cloze deletion and when should I use one?+
A cloze deletion blanks out one element of a sentence and asks you to supply it, leaving the rest visible. Use it when a fact only makes sense in context — a term in a formula, a word in a phrase, a step in a sequence. Blank one element per card, not three.
What should I do with a card I keep forgetting?+
Rewrite it rather than reviewing it harder. A card you fail repeatedly is usually testing two things at once, asking something ambiguous, or phrased in words you never use. After four or five failures, edit the card instead of rating it again.
Put this into practice with knojo.
Turn your notes into a five-minute daily review queue and keep what you learn. Launching soon on iOS and Android.
