Study methods

How to Make Good Flashcards: 10 Rules With Examples

How to make good flashcards that actually stick: ten practical rules, from one fact per card to cloze deletions, with bad-versus-good examples for each.

By the Lernoa team 6 min read

A good flashcard tests one specific fact, asks a real question, and is written in words you’d actually use. Most bad flashcards fail on one of those three counts: they hold too much, ask too little, or copy the textbook. The ten rules below fix that, each with a weak card next to a better one.

Why card quality matters more than card count

Flashcards work because they force you to pull an answer out of memory instead of looking at it. Researchers call this retrieval practice or the testing effect (Roediger and Karpicke, 2006), and in many experiments it has outperformed rereading and highlighting. Dunlosky and colleagues (2013) reached the same ranking when they compared common study techniques. If you want the background, active recall versus rereading covers it in detail.

But retrieval only helps if the card asks for something specific. A vague card gives you a vague feeling of “I sort of knew that”, and you can’t grade a feeling honestly. Every rule below is a way of making the question sharper, so that you know instantly whether you got it right.

Rules 1 to 4: Shape the question

1. One fact per card

If a card has three parts, you will eventually remember two and forget one, and you won’t know which one the scheduler should bring back sooner.

Bad cardBetter cards
Q: Tell me about photosynthesis.Q: Where in the cell does photosynthesis happen? A: In the chloroplasts.
Q: What are the two inputs of photosynthesis? A: Light energy and carbon dioxide plus water.

2. Ask a question, not a topic

A topic on the front (“Mitochondria”) invites you to recite whatever comes to mind. A question demands a specific answer.

Weak front: Mitochondria Better front: What is the main job of the mitochondria?

3. Make the answer short and unambiguous

If two people could give different correct answers, the card is underspecified. Tighten the question until one answer is clearly right.

Weak: What year did the war end? (Which war?) Better: In what year did the First World War end?

4. Avoid lists as answers

A seven-item list is seven facts wearing a trench coat. You’ll recite the first four, feel fine, and never notice the other three are missing. There are better ways to handle a list:

  • Turn each item into its own card (“Which layer of the OSI model handles routing?”).
  • Use a cloze deletion that hides one item at a time.
  • If the order matters, ask for the position: “What is the third step in the process?”

If you truly need the whole set, keep the list short (three or four items) and put the first letters in a mnemonic, which brings us to rule 8.

Rules 5 to 7: Make it yours

5. Use your own words

Copying a textbook sentence onto a card trains you to recognise that sentence, not to understand the idea. Rewrite it the way you’d explain it to a friend. The Feynman technique is built on this: if you can’t explain it simply, you don’t understand it yet, and that’s a signal to go back to the source, not to make a card.

Copied: The process by which a cell’s DNA is replicated semi-conservatively prior to division. Own words: Before a cell divides, it copies its DNA. Each new molecule keeps one old strand and one new strand. Why is this called semi-conservative?

6. Use cloze deletions for sentences and formulas

A cloze card hides a word or phrase inside a sentence, so the context cues the recall. It’s ideal when the surrounding wording matters.

Cloze: In SM-2-style scheduling, pressing “Again” puts the card back into a short {{learning}} step. Formula: The area of a circle is {{πr²}}.

Cloze is weaker for big ideas that need explaining, where a plain question and answer works better.

7. Add an image when the thing is visual

Anatomy, maps, circuit symbols, chart shapes and diagrams are easier to remember as pictures. Put the image on the front and ask for the label, or hide the label and ask for the name. Don’t add decorative images, though: a picture that doesn’t carry information just adds noise.

Rules 8 to 10: Keep the deck healthy

8. Add context or a mnemonic when a fact is arbitrary

Some facts have no logic to hang onto: a port number, a date, a drug name, a French noun’s gender. A one-line hook helps, such as a vivid image, an acronym or a short sentence linking it to something you already know. Put it on the back of the card, under the answer, so it’s there when you check.

Bare cardWith a hook
Q: Which layer of the OSI model handles routing? A: Network (layer 3).Q: Which layer of the OSI model handles routing? A: Network (layer 3). Hook: routers “network” packets between networks.

9. Prune ruthlessly

Some cards turn out to be badly written, too easy, or no longer relevant. Suspend or delete them. A card you’ve failed many times in a row is usually telling you the card is wrong, not that you’re bad at memorising: split it, rewrite it, or add a hook. Leaving leeches in the deck makes every session feel worse and quietly discourages you from coming back.

10. Don’t memorise what you don’t understand

This is the rule people skip. A card can make you recall the words “the Krebs cycle produces NADH” without any idea what that means, and the knowledge won’t transfer to an exam question phrased differently. If you can’t explain the answer, don’t make the card yet. Read, watch or ask until you understand, then create the card to keep the understanding from fading. Flashcards maintain knowledge; they are a poor way to build it from scratch.

A quick checklist before you save a card

Run each new card past these questions:

  1. Does it test exactly one fact?
  2. Could I answer it without seeing the source?
  3. Is there only one correct answer?
  4. Is it in my own words?
  5. If it’s arbitrary, is there a hook?

If any answer is no, edit it now. Ten seconds of fixing saves minutes of frustration across the months you’ll be reviewing it.

Let the scheduler do the remembering

Once your cards are sharp, the next job is seeing them at the right time. A spaced repetition app shows each card just as you’re about to forget it and lengthens the gap when you get it right. That is why good cards pay off so much: each one might be reviewed dozens of times over a year, so a small improvement to its wording is multiplied. If you’re producing cards from lecture material, turning lecture notes into flashcards walks through the workflow.

Generating cards with AI can speed up the first draft. Tools like Lernoa’s AI flashcard generator can build a deck from a topic or a document, but the same rules apply: check that each card is one fact, rewrite the stiff ones and delete anything you don’t understand. Lernoa’s scheduler and review modes are free to use, and you can edit every card it produces.

Try it

Take the ten cards you’re most likely to get wrong this week and rewrite each one using the checklist above. Then review them tomorrow and again a few days later and see which ones still feel fuzzy. If you’d like somewhere to keep them, you can start free and build your deck by hand or from a document.

Frequently asked questions

How many facts should one flashcard contain?

One. If you can get half of a card right and half wrong, the card is really two cards. Splitting it lets a spaced repetition scheduler track each fact separately.

Should I write my own flashcards or use ready-made ones?

Writing your own forces you to process the material, which is part of the learning. Ready-made or AI-generated cards save time, but read them, edit them into your own words and delete the ones you don't understand.

How many flashcards should I review per day?

There is no universal number. Choose a daily limit you can keep up every day, because a steady 20 minutes beats a pile of overdue cards that makes you give up.

Are cloze deletions better than question and answer cards?

Neither is better everywhere. Question and answer cards suit concepts and definitions, while cloze deletions suit sentences, formulas and facts where the surrounding context helps you recall the missing piece.

Keep reading

Study methods

Active Recall vs Rereading: Why Testing Yourself Wins

Active recall beats rereading for long-term memory. See why rereading feels productive, what the testing effect shows, and four ways to practise recall.

6 min read
Learning with AI

How to Use AI to Study Without Cheating Yourself

Learn how to use AI to study: seven practical uses, from quizzes to grounded Q&A, plus the traps to avoid and one simple rule: generate first, then retrieve.

6 min read