MRCPCH flashcards: how to write cards that test recall

Write an MRCPCH flashcard only when you can name the exact knowledge or decision step it should retrieve. Put one unambiguous question on the front, require an answer rather than recognition, keep the checked answer concise, and record the source. Reveal the back only after committing an answer, correct errors immediately, then test whether the knowledge transfers to a fresh single best answer question. A large deck is not evidence of learning; reliable unaided retrieval is.

This guide covers non-clinical examination preparation. It contains no clinical teaching, diagnostic, treatment, prescribing, dosing, safeguarding or patient-management guidance.

The direct answer: design a retrieval event, not a miniature note

A flashcard is useful when its front creates a clear retrieval task and its back provides accurate feedback. It becomes inefficient when the prompt is vague, the answer is a paragraph, the wording gives the answer away or the card tests a detail that will never change a question decision.

The unit of a good flashcard is not a page of content; it is one answerable retrieval decision with a checked source and a reason to exist.

Flashcards can combine retrieval practice with spaced review, but those are separate ideas. Retrieval means attempting to recall before seeing the answer. Spacing means returning after an interval. Simply scrolling through cards, recognising familiar wording or revealing answers immediately may contain little retrieval practice even if the app calls the session "review".

There is no trial showing that a particular flashcard system improves MRCPCH results. Evidence from health-professions education supports distributed and retrieval practice overall, while evidence about voluntary use of electronic flashcards or Anki is more mixed and vulnerable to self-selection. Use cards for bounded retrieval problems and judge them by independent performance, not review counts.

Key points

• Create a card for a specific, verified retrieval problem; do not convert every sentence of a resource into another item to maintain.
• Make the prompt unambiguous, require a committed answer and keep the checked response short enough to score consistently.
• Edit or retire cards that are repeatedly guessed from wording, fail for irrelevant reasons or do not improve recall and fresh-question performance.

What flashcards are good at

Precise facts and distinctions

Cards suit information with a clear cue and a bounded response. They can rehearse a definition, a short comparison, a sequence, a classification boundary or one step in your question process. The answer should be sufficiently precise that you can decide whether recall was correct.

The card does not need to be trivial. It may ask for a short explanation or comparison, but the scoring rule must remain clear. If several paragraphs are required, split the target or use a different learning task.

Repeated retrieval after feedback

The value comes from attempting the answer, checking it and trying again later. A 2014 meta-analysis found that testing previously studied material improved later retention compared with restudy, with larger benefits for initial recall tests than recognition tests. A systematic review in health-professions education included 56 studies and 63 experiments; 43 reported significant benefits from distributed practice, retrieval practice or both.

These findings support the underlying learning activities. They do not show that every deck, card format or scheduling algorithm is effective.

Exposing exact gaps

A well-written card can distinguish "I have seen this" from "I can produce it". When the answer is wrong, the card provides a narrow target for repair. Add the reason the card exists - such as a recurring omission or confused distinction - so later review remains connected to a real problem.

What flashcards are poor at

Replacing representative questions

Current RCPCH guidance describes FOP, TAS and AKP as computer-based single best answer examinations. Flashcards commonly practise cued recall, whereas the examination requires candidates to interpret a stem and discriminate between options. Retrieval practice can transfer beyond the exact practice format, but transfer is not automatic and depends on how practice and later tasks relate.

Cards should therefore support, not replace, timed and untimed MRCPCH question practice, feedback and official sample questions.

Holding an entire topic

Long answers create several scoring problems. You may recall three of seven points and mark the card correct, remember the gist but miss the decisive distinction, or fail one minor item and repeat a large block unnecessarily. A card that behaves like a compressed textbook is hard to retrieve, check and schedule.

Teaching material you have not understood

Do not make a card by copying a sentence you cannot explain. First understand the source, connect the idea to prior knowledge and decide why it matters. Otherwise the deck may train exact wording while leaving the relationship or application unclear.

Measuring readiness

An app's retention percentage, streak or mature-card count is not a validated MRCPCH readiness score. Familiar prompts, repeated wording and self-grading can make deck performance look stronger than performance on first-seen questions. Use broader evidence: delayed recall, mixed questions, error patterns, syllabus coverage and realistic timed work.

What the research says about electronic flashcards

Medical-education evidence is promising but heterogeneous

A 2025 scoping review examined electronic flashcards across health-professions education and highlighted variation in platforms, development and delivery. It called for more systematic study of how design decisions affect learning. A 2026 systematic review of Anki and academic performance found more consistent benefits for standardised licensing examinations than for university-administered examinations, while emphasising limitations in the available evidence.

A separate 2026 systematic review of spaced repetition and retrieval practice in undergraduate medical students included 21 studies. Controlled studies generally favoured structured spaced retrieval, whereas observational studies of self-directed flashcard use were mixed and often at serious risk of bias. Candidates who choose to use many cards may differ from those who do not, so association is not proof that the deck caused the result.

Card design often concentrates on simple recall

A 2026 qualitative analysis classified 1,179 learner-created electronic flashcards from undergraduate medical decks. Cards predominantly rehearsed informational recall and engaged higher-order thinking less often. That does not make simple cards useless. It shows why a deck alone is unlikely to represent all the reasoning, transfer and option discrimination required in an examination.

Retrieval needs correction and transfer

Meta-analytic evidence supports classroom quizzing and retrieval practice, but outcomes vary with feedback, repetition, format and alignment with the final task. A transfer meta-analysis found that testing can support transfer to new contexts, while also showing that the size of transfer depends on features such as response congruency, elaboration and initial performance.

The practical conclusion is modest: use cards to retrieve what cards can test, provide accurate feedback, and add fresh questions to test what cards cannot establish.

Use CARD to create a useful flashcard

CARD is a PaedVault editorial framework, not an RCPCH method or a validated scoring system.

C - Choose one target and one source

Start with a reason, not a document.

Useful reasons include:

• you omitted the same item twice;
• you confuse two neighbouring terms;
• a step in your question process is not automatic;
• delayed recall failed despite understanding; or
• a current official examination detail must be remembered for planning.

Write the source and review date before writing the card. For MRCPCH format, policy or logistics, use a current RCPCH page. For learning claims, prefer peer-reviewed research. If the source changes, the card must be rechecked.

Do not turn every highlighted sentence into a card. Ask: "What future retrieval failure would this prevent?" If there is no specific answer, leave it out.

A - Ask one unambiguous question

The front should specify the answer domain without leaking the answer.

Weak prompt:

What is important about this?

The expected answer is impossible to score because the cue has no boundary.

Stronger prompt shape:

What are the two criteria I use before moving a revision topic from repair to mixed practice?

This defines the number and context. It tests a personal process, not clinical content, and has a stable scoring rule.

Avoid prompts that can be answered only because of deck order, image position, unusual punctuation or a conspicuous word on the front. Shuffle cards and occasionally rewrite the cue. If accuracy collapses, the original card may have trained a surface signal rather than the target.

R - Retrieve, reveal and repair

Read the front and produce an answer before revealing the back. Say it, type it or write it. Mental familiarity is not enough when the target requires production.

Then compare the answer with the checked response. Use a consistent grading rule:

• correct: the complete essential answer was retrieved;
• partial: the main idea appeared but a required element was missing or uncertain;
• incorrect: the answer was absent, wrong or guessed from an irrelevant cue.

Do not repeatedly mark partial answers as correct to protect a streak. Repair the cause. If the answer is too broad to grade, rewrite the card. If the source was misunderstood, return to it. If the prompt is ambiguous, fix the front rather than blaming memory.

The self-explanation guide offers a fuller process for making reasoning visible and repairing errors.

D - Demonstrate transfer or delete the card

A successful card review shows that you answered that cue. It does not yet show that you can recognise the relevant knowledge in a different stem or distinguish a strong alternative.

After several successful reviews:

1. explain the answer without the original wording;
2. answer the reverse or contrasting question;
3. use a fresh single best answer item;
4. retrieve it after a longer delay; and
5. decide whether the card still earns its maintenance cost.

Retire cards that are now directly retrievable, duplicate a better card, test obsolete information or never influence question performance. A smaller deck with a job is better than an archive that grows faster than it can be checked.

Five card-design decisions that matter

1. Basic question-and-answer or cloze deletion?

A basic card states a question on the front and an answer on the back. It is usually easier to make the expected task explicit.

A cloze card removes a word or phrase from a sentence. It is fast to create, but surrounding wording can make the missing text obvious without requiring the intended knowledge. Cloze cards work best when the sentence is understood, the blank tests a meaningful distinction and the remaining text does not act as an accidental answer key.

Choose the form that creates the cleanest retrieval task. No format is automatically more advanced.

2. One item or a short related set?

"One fact per card" is a useful warning against paragraphs, but it is not an absolute law. Some targets are meaningful only as a short set or contrast. Keep items together when the relationship is the knowledge; split them when one forgotten element forces unnecessary review of several known elements.

Write the required number on the front. If the answer contains four elements, the cue should say four. That makes self-grading more reliable.

3. Recognition or recall?

Multiple-choice cards can rehearse discrimination, but repeated options can become familiar. Open prompts usually require more production. The testing-effect meta-analysis found larger benefits for initial recall than recognition in the studies it reviewed.

For a factual target, attempt open recall first. Use representative single best answer questions separately to practise option comparison. Do not turn your whole deck into repeated miniature MCQs and assume it recreates first-seen exam performance.

4. Text, image or both?

Use an image only when the visual itself is the retrieval target or clarifies structure. Decorative screenshots increase clutter and may contain hidden cues. Ensure that you are allowed to use the image, record its source and keep essential scoring information accessible in text.

This article does not recommend clinical images or teach their interpretation. Follow current RCPCH material and appropriately governed resources for examination image practice.

5. Your own cards or a shared deck?

Creating a card can require selection and explanation, but card creation also consumes time. A shared deck saves construction time but may contain outdated, irrelevant, ambiguous or incorrect items. Research has not established a universal winner for MRCPCH candidates.

If you import cards, quarantine them before review. Check each card against your current source, rewrite the prompt in language you understand, remove irrelevant material and tag the reason for keeping it. Never import a large deck directly into the daily queue merely because it is popular.

How to repair common bad cards

The encyclopaedia card

Symptom: the answer is several paragraphs or a long screenshot.

Repair: identify the one distinction that caused an error. Create one or two bounded prompts and use a mind map or concept map for the larger relationship.

The vague card

Symptom: several answers could reasonably fit the front.

Repair: add the context, direction and number of required elements. If the cue becomes longer than the knowledge, use a different method.

The giveaway card

Symptom: grammar, layout, deck order or surrounding words reveal the answer.

Repair: rewrite and shuffle it. Test the card outside its original deck. Make the target retrievable from meaning rather than position.

The orphan card

Symptom: you can recite the answer but cannot explain why it matters or where it came from.

Repair: add a source and a short "why this exists" note. Connect it to a question error, syllabus point or defined study decision. Delete it if no useful connection exists.

The immortal card

Symptom: an easy item remains in frequent review because the algorithm or your settings never allow it to leave.

Repair: check scheduling settings, lengthen intervals or retire it. The separate guide to spaced repetition without flashcard overload explains how to manage review volume and leeches.

The broken card blamed on memory

Symptom: the item fails repeatedly for different reasons.

Repair: inspect the design before adding more repetitions. The target may be too broad, the source unclear, the wording unstable or the card testing several decisions at once. Relearn, split, rewrite or delete.

Build cards from question-bank errors without copying questions

Do not reproduce proprietary question stems, explanations or recalled examination content. Extract your learning need in your own words from a source you are permitted to use.

A safe workflow is:

1. finish the question and feedback process;
2. classify the error;
3. identify the smallest verified retrieval failure;
4. write a new prompt without copying the stem or options;
5. cite the authoritative source rather than the question itself; and
6. test transfer later with a fresh question.

Not every wrong answer needs a card. Misreading, pacing, confidence or premature commitment may require a process change rather than another fact in the queue.

Use AI-generated cards cautiously

Generative AI can draft many cards quickly, but volume magnifies checking costs. It may invent facts, cite nonexistent sources, create ambiguous cloze deletions or select details that do not match the MRCPCH syllabus.

If you use AI, provide only material you are authorised to share, require it not to add facts, and inspect every card against the source before import. Ask for a small sample first. Delete cards that you cannot personally verify and score. The safe AI workflow for MRCPCH revision explains verification and privacy boundaries in more detail.

Automatic generation does not remove the maintenance cost. A hundred unchecked cards are a hundred possible errors scheduled for repetition.

A weekly flashcard audit

Once a week, sample rather than admire the dashboard.

Check:

• five recently added cards for source accuracy;
• five repeatedly failed cards for design faults;
• five mature cards using changed wording;
• one mixed question block for transfer;
• the number of new cards versus retired cards; and
• total review time versus your planned question and feedback time.

Keep one decision per card: retain, rewrite, relearn, suspend or delete. Add the results to your weekly MRCPCH revision review. If the queue keeps growing while mixed-question performance is flat, stop adding cards until the mismatch is understood.

Conclusion

Effective MRCPCH flashcards are deliberately small, verifiable retrieval tasks. Use CARD: choose one target and source, ask an unambiguous question, retrieve before revealing and repair errors, then demonstrate transfer or delete the card.

Flashcards are one component of a broader preparation system. They can strengthen access to selected knowledge, but they cannot establish broad syllabus coverage, realistic pacing or single best answer discrimination on their own. Measure the deck by what becomes independently retrievable and usable - not by how many cards exist.

Frequently asked questions

Are flashcards effective for MRCPCH revision?

They can support retrieval and spaced practice for bounded knowledge, but no study shows that flashcards improve MRCPCH pass rates. Health-professions evidence supports distributed and retrieval practice overall, while observational evidence about self-directed electronic flashcards is mixed. Use cards alongside verified sources, feedback and fresh single best answer questions.

How many MRCPCH flashcards should I make?

There is no evidence-based target. Create a card only for a verified retrieval problem that matters, and limit new cards to a volume you can review without displacing question practice, feedback or sleep. Audit and retire cards regularly; deck size is not a readiness measure.

Should I make my own flashcards or use a premade deck?

Neither is universally superior. Making cards can force selection and explanation but costs time. A premade deck saves construction time but may be inaccurate, outdated or poorly matched to the MRCPCH syllabus. Verify every imported card, rewrite unclear prompts and add only cards that solve a specific learning problem.

Should an MRCPCH flashcard contain one fact?

Usually keep one clearly scorable retrieval target per card. A short set or comparison can remain together when the relationship is the knowledge, but state the required number and avoid paragraph-length answers. Split a card when one missed element forces repeated review of several items already known.

Can flashcards replace MRCPCH question-bank practice?

No. Flashcards usually practise cued recall, while MRCPCH theory examinations use single best answer questions that require interpretation and option discrimination. Use cards to repair selected retrieval gaps, then confirm transfer with first-seen questions, feedback and timed blocks.

Sources

1. Royal College of Paediatrics and Child Health. Theory exams: structure and syllabi. Updated 27 August 2026.
2. Trumble E, et al. Systematic review of distributed practice and retrieval practice in health professions education. Advances in Health Sciences Education. 2024.
3. Rowland CA. The effect of testing versus restudy on retention: a meta-analytic review of the testing effect. Psychological Bulletin. 2014.
4. Pan SC, Rickard TC. Transfer of test-enhanced learning: meta-analytic review and synthesis. Psychological Bulletin. 2018.
5. Yang C, et al. Testing (quizzing) boosts classroom learning: a systematic and meta-analytic review. Psychological Bulletin. 2021.
6. Electronic flashcards in health professions education: a scoping review. Academic Medicine. 2025.
7. Anki use and academic performance in medical education: a systematic review of evidence and learning theory. 2026.
8. A systematic review of spaced repetition, retrieval practice and academic performance in undergraduate medical students. 2026.
9. Design of learner-created electronic flashcards in undergraduate medical education. 2026.