What is a good MRCPCH question bank score?

There is no validated MRCPCH question-bank percentage that guarantees a pass. A useful score is one produced from first-seen questions, broad syllabus coverage and reasonably consistent conditions, then interpreted as a trend rather than a verdict. Track why answers were wrong and whether corrected distinctions survive a delayed retest. Do not treat a platform percentage as the RCPCH pass mark.

Candidates understandably want a number. A dashboard shows 54%, 67% or 78%, and the immediate question is: Is that good enough? The percentage looks precise, but its meaning changes with the questions selected, whether they are new or repeated, whether sources were open, how recently the topic was revised and how many items were attempted.

The better question is not "What score should everybody reach?" It is "What does this score allow me to decide about my next block of revision?"

Why there is no universal good MRCPCH question-bank score

An informal question-bank result and an official examination result are different measurements.

RCPCH sets each MRCPCH theory examination standard using a criterion-referenced Modified Angoff process. Subject-matter experts judge items against the expected borderline candidate, and the resulting item judgements contribute to the pass mark. RCPCH states that the pass mark and pass rate may vary between diets because the difficulty of the paper is considered. From the 2025.2 diet, the judging process uses two rounds of judgement, with the final judgements averaged for each question.

That means a fixed percentage copied from a forum, another candidate or a commercial dashboard is not the official standard for your future sitting. A question bank has its own item mix, difficulty distribution, repetition rules and user population. Unless a provider has published appropriate validation evidence for a specific prediction, its raw percentage should not be presented as a pass-equivalent score.

RCPCH also warns candidates not to average the category percentages on their official feedback to reconstruct the overall examination score. Categories can contain different numbers of questions, and some questions cover more than one category. If official category feedback cannot be combined by simple averaging, an informal collection of question-bank percentages deserves at least as much caution.

The PaedVault MRCPCH guide summarises the current examination pathway and links to the official pages that candidates should recheck near their sitting.

A question-bank percentage is useful when it is a comparable sample of your current retrieval. It becomes misleading when it is treated as a universal pass mark.

What can change the meaning of the same percentage?

Imagine two candidates both record 70%.

Candidate A answered 100 first-seen questions sampled across the syllabus, without notes, in several timed blocks. Candidate B answered 20 questions from a favourite topic immediately after revising it, recognised several repeats and checked two answers before submitting.

The displayed percentage is identical. The evidence is not.

Seven common factors change interpretation:

• First-seen or repeated: a repeated item can measure memory for the item, its wording or its answer as well as the underlying distinction.
• Coverage: a narrow, familiar topic can hide omissions elsewhere in the syllabus.
• Question selection: random, targeted and self-selected sets answer different questions about performance.
• Support: open notes, discussion and source checking turn an attempt into guided learning rather than an independent sample.
• Timing: untimed learning blocks and examination-paced blocks serve different purposes.
• Sample size: in a 20-question block, one answer changes the percentage by five points. A short block can therefore look dramatically better or worse after only a few items.
• Recency: performance immediately after study may not survive a delay.

Testing specialists distinguish reliability - how consistently scores measure something - from validity, which asks whether the score supports the intended interpretation. A stable score on repeated familiar questions could be reliable for remembering those questions but invalid as evidence of broad first-seen readiness. An occasional difficult block can be a useful learning event even when its percentage is low.

Use a four-lens MRCPCH scorecard

Do not discard accuracy. Put it beside three other signals so that one percentage cannot dominate the decision.

Lens	Record	Question it answers
Comparable accuracy	First-seen correct answers under labelled conditions	Is independent performance changing across similar blocks?
Coverage	Syllabus areas represented and omitted	Is the score broad or concentrated in familiar territory?
Error structure	Knowledge, interpretation, comparison, timing or avoidable-process errors	What should change next?
Delayed retention	Independent result on corrected distinctions after a delay	Did review produce retrievable learning?

This is not a validated MRCPCH prediction model. It is a revision-control system: evidence enters, a study decision follows, and later retrieval checks whether that decision worked.

Lens 1: comparable first-seen accuracy

Keep a separate series for questions you have not previously attempted. Label the conditions that materially affect comparison:

• exam component or scope;
• random or targeted selection;
• timed or untimed;
• closed-source or open-source; and
• number of questions.

Compare like with like. A targeted untimed learning block should not be placed on the same trend line as a broad, timed, closed-source block. Both are legitimate; they answer different questions.

Use a rolling view across several comparable blocks instead of reacting to a personal best or worst. Test scores can vary because a different sample of questions happens to align differently with what a learner knows. The ETS guide to test reliability describes question selection, occasion and testing conditions as possible sources of score inconsistency. An informal bank does not give you enough evidence to calculate away all that uncertainty, so the practical response is modest: gather more comparable observations before making a large decision.

Lens 2: coverage

A rising percentage can be genuine and still conceal a narrow study pattern.

Create a simple coverage map for the examination scope you are preparing for. Mark an area only when it has contributed first-seen questions under a defined condition; do not mark it merely because you read notes about it. Then ask:

1. Which areas have recent independent evidence?
2. Which areas have only supported or repeated attempts?
3. Which areas have no recent sample at all?

The aim is not to force equal time into every label. It is to prevent the dashboard from being driven by repeated success in a small, comfortable corner. The interleaving guide explains when mixed practice can test discrimination and when a short blocked phase may still be useful.

Lens 3: error structure

"Wrong" is a result, not a revision instruction. Give each consequential error one primary cause:

• knowledge gap: the required fact or rule was not retrievable;
• unstable comparison: two plausible alternatives were not distinguished;
• task misread: the stem asked for something different from the answer supplied;
• unsupported inference: the reasoning added information that was not present;
• timing: the decision process was not completed within the intended pace; or
• avoidable process: an answer was changed, omitted or entered contrary to the candidate's own settled reasoning.

The categories are for reflection, not official RCPCH reporting. Keep them few enough to use consistently.

A block at 72% with ten versions of the same unstable comparison may require a different response from a block at 72% containing scattered, low-confidence gaps. The first suggests a focused discrimination exercise. The second may call for broader retrieval and source-guided repair.

When reviewing, use self-explanation to state why the preferred option is better, and use the feedback-timing guide to decide when to open the explanation. This article does not provide clinical teaching; substantive content should be checked against the referenced explanation and appropriate authoritative source.

Lens 4: delayed retention

Immediate correction can create a persuasive sense of mastery. Close the explanation and retrieve the corrected distinction later, without copying the original wording.

Possible checks include:

• answer a changed question that requires the same distinction;
• state the decision rule from a blank prompt;
• explain why the previously attractive alternative is weaker; or
• reattempt the original only after enough delay that simple visual familiarity is less helpful.

Health-professions and classroom reviews generally support retrieval practice over restudy for later learning, though interventions and effect sizes vary. Feedback is valuable, but its impact depends on the information it provides and what the learner does next. The scorecard therefore treats review as incomplete until a later independent retrieval checks the result.

The spaced-repetition guide offers a bounded way to schedule those checks without turning every explanation into a permanent flashcard.

Build a score you can actually compare

Use the SCORE labelling rule before starting a block.

S - Scope

State what the block samples: FOP, TAS, AKP, a named syllabus area or a mixed set. A score without scope cannot show whether coverage is widening.

C - Conditions

Choose timed or untimed and closed-source or open-source. If the conditions change, start a different series instead of pretending the numbers are comparable.

O - Originality

Separate first-seen questions from repeats. Repeats can test correction and retention, but they should not inflate the first-seen trend.

R - Reasons

Classify the few errors that will change the next study action. Do not produce an elaborate taxonomy that takes longer than the repair.

E - Evidence later

Schedule a delayed, individual retrieval for corrected distinctions. Record whether the answer and reasoning survive.

SCORE turns "I got 68%" into "I got 68% on a broad, first-seen, timed, closed-source set; most losses came from two unstable comparisons, which I will retest after repair." The second statement can guide revision. The first mostly invites anxiety or reassurance.

A practical weekly dashboard

One row per comparable block is enough:

Date	Scope	First-seen?	Conditions	Correct/total	Main error pattern	Delayed check due
Monday	mixed scope	yes	untimed, closed source	14/20	unstable comparisons	Thursday
Wednesday	targeted area	yes	untimed, closed source	16/20	knowledge gaps	Saturday
Friday	mixed scope	yes	timed, closed source	15/20	task misreads	Tuesday

These figures are examples, not target scores. The rows should remain in separate comparison series when scope or conditions differ.

At the end of the week, make only a few decisions:

1. Continue: the next comparable block should test whether the trend persists.
2. Repair: one repeated error pattern needs a short source-guided intervention.
3. Broaden: neglected areas need first-seen sampling.
4. Retest: corrected distinctions are due for delayed retrieval.

Monitoring becomes useful when it changes behaviour. A 2022 meta-analysis found moderate positive effects of self-monitoring interventions on strategy use and academic performance across included educational studies, with stronger findings when monitoring sat inside a broader intervention and received environmental support. That evidence is not MRCPCH-specific, but it supports using a compact dashboard as part of an action cycle rather than collecting numbers for their own sake.

Avoid six misleading score habits

Chasing a crowd-sourced pass percentage

Another candidate's number may come from another bank, item mix, stage of revision or use of repeats. It cannot replace current RCPCH standard-setting information or validated prediction evidence.

Combining first attempts and repeats

This hides whether improvement reflects broader retrieval or memory for previously seen material. Keep two series.

Resetting the bank to erase weak performance

Deleting the trace does not remove the gap. Preserve enough history to see which repairs survived.

Selecting only comfortable topics

The percentage may rise while readiness becomes less representative. Pair targeted repair with later broad sampling.

Treating every fall as regression

A more difficult or broader sample can lower accuracy while providing better information. Check the conditions and error pattern before changing the whole plan.

Treating every rise as mastery

Recent exposure, hints, repeats and open sources can lift immediate accuracy. Ask whether the same distinction survives a delayed, unsupported attempt.

The confidence-calibration guide adds a complementary signal: confidence attached to each answer can reveal correct guesses and high-confidence errors that percentage alone conceals.

When should a question-bank score change your plan?

Use a score to make a small, reversible change when the evidence is comparable and repeated.

• A persistent error cluster can trigger a focused repair block.
• A coverage gap can trigger first-seen sampling from an omitted area.
• A stable broad trend can justify adding more timed conditions.
• Failed delayed checks can shorten the interval to the next retrieval or simplify the material being retained.
• A single unusual block should usually trigger inspection before a major timetable rewrite.

Do not use a private question-bank score to declare that you will pass or fail. Official MRCPCH results come from the examination and its current standard-setting process. If you want to rehearse performance, keep learning blocks distinct from formal mock conditions and use multiple observations rather than one dramatic percentage.

How PaedVault can support honest score interpretation

PaedVault sessions can be used for different jobs, but name the job before you begin:

• learn: targeted, untimed work with deliberate feedback;
• sample: broader first-seen questions under closed-source conditions;
• retest: delayed checks of corrected distinctions; or
• rehearse: a longer, timed session designed to practise pacing and sustained attention.

Do not demand that one session perform all four jobs. A learning session can be successful precisely because it exposes gaps. A sampling session should preserve independence. A retest should make previous corrections retrievable. A rehearsal should be judged partly by process, not just the final percentage.

If discussion helps resolve competing reasoning, the MRCPCH study-group guide shows how to place collaboration between two individual attempts so that collective accuracy does not replace personal recall.

Conclusion

There is no universal good MRCPCH question-bank percentage and no responsible conversion from a raw platform score to a future RCPCH result without appropriate validation.

Keep accuracy, but make it interpretable. Separate first-seen questions from repeats, label conditions, check syllabus coverage, classify useful error patterns and verify corrections after a delay. Judge several comparable blocks together. The goal is not to manufacture certainty from a dashboard; it is to make the next revision decision better.

Frequently asked questions

What percentage should I get on an MRCPCH question bank?

There is no validated universal percentage that guarantees an MRCPCH pass. Question banks differ in difficulty, selection, repetition and conditions, while RCPCH sets examination pass marks using a criterion-referenced process that can vary by diet. Use comparable first-seen trends, coverage, error patterns and delayed retention rather than one threshold.

Does a high PaedVault score mean I am ready for MRCPCH?

It is encouraging only in context. Check whether the score came from broad first-seen questions under independent conditions and whether performance is stable across several blocks. Repeats, narrow topic selection, recent revision and open sources can make a high percentage less representative. PaedVault does not claim that one raw percentage guarantees an official result.

Should I reset my MRCPCH question-bank score?

Usually keep enough history to distinguish first attempts from repeats and to see trends. Resetting can hide useful evidence. If you start a new phase, label it as a separate series with its scope and conditions rather than deleting earlier performance.

How often should I check my MRCPCH question-bank average?

Review it after several comparable blocks, not after every answer. There is no validated universal interval. A weekly review can be practical if it leads to a small decision about coverage, error repair, timed practice or delayed retesting rather than repeated reassurance-seeking.

Sources

1. Royal College of Paediatrics and Child Health. Education and training standard setting.
2. Royal College of Paediatrics and Child Health. Amendment to the theory exam standard setting process in 2025.
3. Royal College of Paediatrics and Child Health. Theory exams - what to expect after your exam.
4. Livingston SA. Test Reliability - Basic Concepts. ETS Research Memorandum RM-18-01. 2018.
5. Green ML, Moeller JJ, Spak JM. Test-enhanced learning in health professions education: a systematic review. Medical Teacher. 2018;40(4):337-350.
6. Phillips JL, Heneka N, Bhattarai P, et al. Systematic review of distributed practice and retrieval practice in health professions education. Advances in Health Sciences Education. 2024;29:689-714.
7. Guo L. The effects of self-monitoring on strategy use and academic performance: a meta-analysis. International Journal of Educational Research. 2022;112:101939.
8. Wisniewski B, Zierer K, Hattie J. The power of feedback revisited: a meta-analysis of educational feedback research. Frontiers in Psychology. 2020;10:3087.