To make a good flashcard, put one question on the front that forces you to produce an answer, and put only that answer (plus a one-line reason) on the back. Then study by answering out loud or on paper before you flip, shuffle the whole deck each pass, and keep a card in rotation until you've got it right in more than one session.
Most advice stops at "one fact per card," which works for vocabulary but not for math, chemistry or physics. This guide rewrites real STEM cards from weak to strong, explains the research behind each change, and ends with a review schedule you can copy.
The One Rule of a Good Flashcard
If you can read the front and think "yeah, I know that" without actually saying the answer, the card is a summary, not a test. That's a common reason a deck of 200 cards fails a student on exam day.
Why Flashcards Work (When You Use Them Right)
The idea is called retrieval practice, and it has strong evidence behind it. In Roediger and Karpicke's 2006 study, students who reread a passage did better on a test five minutes later, but students who practiced recalling it did better two days and one week later. Rereading also made students more confident in the same study, even though it left them remembering less, which is exactly why rereading is so tempting.
A large review of ten study techniques by Dunlosky and colleagues (2013) gave its highest rating to two of them, practice testing and distributed practice (spreading study out over time), and rated rereading and highlighting low. A well-used flashcard deck does both of the high-rated things at once.
The deck also needs spacing. In Kornell's 2009 flashcard experiments, spaced study beat cramming for 90% of participants, yet after the first session 72% of them believed cramming had worked better. The same experiments found that one big stack beat splitting the same cards into small stacks, because a big stack puts more time between repeats of each card.
How to Make a Flashcard in 5 Steps
- Understand it first
Cards are for remembering, not for learning something from zero. Work the textbook section or class notes until the idea makes sense, then make cards from it.
- Pick one target per card
Ask: what exactly should I be able to produce? A formula, one step, one reason or one choice. If the back needs two separate marks to grade, split it into two cards.
- Write the front as a question
"Quadratic formula" is a topic. "Solve
ax² + bx + c = 0for x" is a question. Questions force retrieval; topics invite a nod. - Keep the back short, with the why
The answer, its unit if it has one, and one line saying why. The why line is what makes the card useful on a problem you've never seen.
- Write it in your own words
A copied textbook sentence can feel familiar later even if you never understood it. Rephrasing makes you decide what the sentence actually means.
Before and After: Rewriting Real Flashcards
Each example starts with a card students really make, says what is wrong with it, and rebuilds it.
Before. Front: Unit circle. Back: a full copied circle with 16 angles and their coordinates.
Problem: the front asks nothing, and the back has 32 values. You can't mark it right or wrong, so you just look at it.
After. Front: sin(5π/6) = ? Back: 1/2. Why: reference angle π/6, Quadrant II, where sine is positive.
Make a few of these per quadrant rather than one per angle. The why line trains the method from the unit circle guide (reference angle plus sign), so it works for angles you never put on a card.
sin(5π/6) = 1/2Before. Front: Quadratic formula. Back: the formula, what the discriminant means, and a worked example, all in one paragraph.
Problem: three targets on one card. You'll remember the formula, feel good, and never notice you forgot the discriminant.
Card A. Front: Solve ax² + bx + c = 0 for x. Back: x = (−b ± √(b² − 4ac)) / 2a. Why: it is what completing the square gives for any quadratic.
Card B. Front: b² − 4ac < 0. How many real solutions? Back: none (two complex solutions), because you can't take the real square root of a negative number.
Card C. Front: Solve 2x² + 3x − 2 = 0. Back: b² − 4ac = 9 + 16 = 25, so x = (−3 ± 5) / 4, giving x = 1/2 or x = −2.
Card C is a full problem, so work it on paper before flipping. If you keep missing one step of it (say, the sign of −4ac), give that step its own card. Change the numbers each time you rewrite the deck, and check a new version with the quadratic formula calculator, which shows its working.
x = 1/2 or x = −2Before. Front: Mole. Back: the textbook definition of the mole, copied word for word.
Problem: you can recite that sentence and still not convert grams to moles, which is what tests ask for.
After. Front: How many moles are in 36.0 g of H₂O?
Back: n = m / M. Molar mass M = 2(1.008) + 15.999 = 18.015 g/mol, using the IUPAC standard atomic weights (H 1.008, O 15.999).
n = 36.0 g ÷ 18.015 g/mol = 1.998 mol. The grams cancel, leaving mol. The data has 3 significant figures, so round to 2.00 mol.
n = 2.00 molBefore. Front: Kinematic equations. Back: all five equations in a list.
Problem: on a test nobody asks you to list the equations. They ask you to pick the right one. This card never practices the choice.
After. Front: A car starts from rest and accelerates at 3.0 m/s² for 4.0 s. Which equation gives the distance, and what is it?
Back: known u = 0, a = 3.0 m/s², t = 4.0 s; wanted s; final speed not given or needed. So use s = ut + ½at².
s = 0 + ½ × 3.0 m/s² × (4.0 s)² = ½ × 3.0 × 16 m = 24 m. Units: m/s² × s² = m.
The skill being tested is the list of knowns and the one missing variable, which is the method in the kinematic equations guide. Write a few cards like this where the "missing variable" changes each time.
s = ut + ½at², s = 24 mWhat to Put on the Front and Back
The four examples use different card types. A good STEM deck mixes them, because each one trains a different thing a test asks for.
| Card type | Front | Back |
|---|---|---|
| Fact | sin(5π/6) = ? | Value plus the one-line reason |
| Formula | "Solve ax² + bx + c = 0 for x" | The formula and where it comes from |
| Why | "Why can't b² − 4ac < 0 give real roots?" | One sentence of reasoning |
| Cloze (fill the gap) | "In n = m / M, M is the ____, in g/mol." | Molar mass |
| Worked problem | A short problem with numbers | Key steps and the answer, with units |
| Method choice | A situation: "which equation or rule?" | The choice, why, then the result |
Keep worked-problem cards short enough to finish in about a minute. Anything longer belongs in a practice set, not a deck.
Flipping the card the moment you "sort of" know the answer. Recognizing the back when you see it feels the same as knowing it, but it isn't. Say the answer out loud or write it down first, then flip and mark yourself honestly: a half-right answer is a miss.
How to Study Your Flashcards: A Schedule You Can Copy
Making the cards is half the job. The other half is deciding when each card comes back. Two findings shape the schedule below.
First, don't retire a card too early. In Kornell and Bjork's 2008 experiments, letting learners drop the flashcards they thought they knew had small but consistently negative effects on learning. Second, in a 2022 review, Rawson and Dunlosky describe successive relearning, which combines retrieval practice with spaced practice: get each item right in a first session, then come back and recall it again in later, spaced sessions.
A three-box system (a simple form of spaced repetition) builds both in. Use three envelopes, rubber-banded piles or the decks in an app:
- Box 1 (new and missed cards): review every day.
- Box 2: review every 3rd day (days 3, 6, 9, 12, ...).
- Box 3: review every 7th day (days 7, 14, 21, ...).
A correct answer moves a card up one box. A miss sends it back to Box 1, wherever it was. Shuffle each box before you go through it.
You make Card C (the quadratic) on day 1 and get it right every time it appears. When do you see it?
Day 1: Box 1, correct, moves to Box 2.
Day 3: first Box 2 session, correct, moves to Box 3.
Day 7: first Box 3 session, correct, stays in Box 3. Day 14: correct again.
The gaps between reviews are 2, 4 and 7 days. They grow because each success earns a longer wait. If you miss it on day 7, it drops to Box 1 and shows up on day 8.
Days 1, 3, 7 and 14Over those 14 days you open Box 1 on 14 days, Box 2 on 4 days (3, 6, 9, 12) and Box 3 on 2 days (7, 14). Most days are a short Box 1 session, which is easier to keep up than one long weekend cram.
How many flashcards a day?
No study gives a single right number, so set it by your review load. Every new card comes back: the card in Example 5 is seen 4 times in two weeks even when you never miss it. Ten new cards on day 1 therefore mean at least 40 reviews by day 14, and more for every card you miss. Add new cards only as fast as you can still clear Box 1 in one sitting, and stop adding for a day or two if it starts to pile up.
Paper Cards or an App?
Both work. In practice, how you use the cards matters more than what they are made of. Paper makes it easy to draw a graph, a molecule or a free-body diagram on a card, and to color-code by topic (say, blue for formulas, red for method-choice cards). Many flashcard apps handle the schedule for you and make it painless to keep hundreds of cards in one shuffled deck, which suits the big-stack finding above.
Solver AI's flashcards and quizzes feature can generate a deck in three modes (flashcards, multiple choice and true/false) for revising for tests such as the SAT, ACT, AP or GCSE; treat AI-made cards like a classmate's, checking each one against your notes and splitting any that test two things. If a worked-problem card stumps you, you can scan it with the app to see the steps on your iPhone or iPad, then rewrite the card around the step you missed.
Whatever you use, cards are for study time. Phones and apps stay out of the exam room unless the exam's own rules say otherwise.
How to Tell It Is Working
The feeling of fluency while flipping cards is not good evidence, as the confidence result in the 2006 study shows. Use checks that look like the real test instead:
- Track first-try accuracy
Count how many Box 2 and Box 3 cards you get right on the first attempt each session. That number should rise over a week. If it doesn't, the cards may be too big: split them.
- Do unseen problems
Once a week, try a few end-of-chapter problems you haven't put on cards. Method-choice cards should make these feel familiar. If not, add more of that card type.
- Blank-page recall
Close the deck and write everything you know about one topic on a blank sheet. Gaps there point to missing cards.
One question per card, an answer you produce before flipping, one big shuffled deck, and spaced sessions where a card earns its way to longer gaps. Flashcards then do the two things the Dunlosky review rated highest: testing yourself and spreading it out.
Try It: Build Your First Deck
Work through these in order. Answers follow each task so you can compare.
- Split a crowded card
Rewrite this card as two: "Newton's second law: F = ma, force equals mass times acceleration, measured in newtons." One answer: Card 1, "Net force on a 2.0 kg cart accelerating at 3.0 m/s²?", back
F = ma = 2.0 kg × 3.0 m/s² = 6.0 N. Card 2, "What is 1 N in base SI units?", back1 kg·m/s². - Make a cloze card
Turn
n = m / Minto a fill-the-gap card for the unit of M. One answer: "M inn = m / Mis measured in ____." Back: g/mol. - Make a method-choice card
Write the back for: "A ball is dropped from rest from 20 m. Which equation gives the time to fall, and what is it? (g = 9.8 m/s²)". Answer: knowns
u = 0,s = 20 m,a = 9.8 m/s², want t, sos = ut + ½at²givest = √(2s/a) = √(40 / 9.8) s ≈ 2.0 s. - Plan three weeks
Using the three-box schedule, list the days in a 21-day plan when you open Box 2 and Box 3, and when a card made on day 1 and always answered correctly comes back. Answer: Box 2 on days 3, 6, 9, 12, 15, 18, 21 (7 sessions); Box 3 on days 7, 14, 21 (3 sessions); the card returns on days 3, 7, 14 and 21.