This audit is for the core psychiatry trainee who already uses, or is about to buy, MRCPsychMentor for MRCPsych Paper B and wants to read its analytics honestly rather than be reassured by them. It addresses the written MCQ-and-EMI paper — clinical topics plus the substantial critical-review and statistics component — not the CASC. The principal limitation is the same as for any bank: the platform scores the questions you chose to attempt, in the modes you chose, so the home-screen percentage is a study metric, not an exam-day estimate.
What MRCPsychMentor offers for MRCPsych Paper B right now
The figures below are vendor-reported, taken from the MRCPsychMentor Paper B pages as observed on 20 July 2026. Treat every count, price and feature as provisional and confirm the current numbers on the product page before you buy.
| Item | Vendor-reported detail (20 July 2026) |
|---|---|
| Bank size | "Over 2,000" questions, curated from previous exams and the current College syllabus |
| Question format | Best-answer one-of-five MCQs and extended matching items (EMIs), matching the written paper |
| Domains | Sections 6–14 of the syllabus: organisation and delivery of services; general adult; old age; psychotherapy; child and adolescent; substance misuse; forensic; learning disability; and research methods, statistics, critical review and evidence-based practice |
| Analytics | A histogram comparing your scores against other candidates, updated daily; strength-and-weakness analysis; powerful question-review functions — no published national percentile or predicted pass score |
| Difficulty control | Candidate-selected topic and mode (revision mode versus timed test) |
| Mock exams | Two mock exams structured to match the distribution of questions by category used by the College |
| Explanations | Answer explanations and revision notes under each question; flagging and annotation |
| Access and price | 4 months £60; 6 months £80 — verify current pricing on the product page |
| Trial | Free demo |
| Exam components | Written Paper B MCQ/EMI; a separate AI-powered CASC resource exists but is not part of the Paper B bank |
One point governs how you should read the rest of this article. Like most banks marketed as "intelligent", MRCPsychMentor does not, on the pages we checked, publish or claim a proprietary adaptive-difficulty algorithm, a national percentile or a predicted pass score. What it offers is candidate-selected topic practice, a daily-updated peer-comparison histogram, strength-and-weakness identification and two category-weighted mocks. This audit treats the product as what it is and makes no proprietary-algorithm claims; verify the current feature language yourself rather than assuming the feed adapts difficulty for you.
The exam this dashboard is trying to predict
MRCPsych Paper B is a single computer-delivered paper: 150 questions in three hours at a Pearson VUE centre, approximately two-thirds best-of-five MCQ and one-third EMI. Its content splits in a way that shapes every study decision: one third of the paper is critical review — research methods, statistics, critical appraisal and evidence-based practice, roughly 50 marks of the 150 — and two thirds is clinical, spread across general adult, old age, child and adolescent, forensic, addictions, psychotherapy, learning disability and the organisation of services. There is no negative marking, and the pass mark is set by a criterion-referenced standard (confirm current standard-setting on rcpsych.ac.uk). Paper B is a written gateway; the clinical CASC follows as a separate OSCE-style exam that neither an MCQ bank nor iatroX reproduces.
Two things follow. First, the College syllabus, published sample questions and reading list are the calibration standard — a vendor bank is a training tool, not the blueprint. Second, the fixed one-third critical-review weighting means statistics and appraisal are not optional garnish; they are a third of your marks, and the single most common Paper B study error is treating them as such.
Every metric on the dashboard, defined
You cannot audit a number you have not defined. Here is what each metric means, and what it does not.
First-attempt accuracy is the percentage you answered correctly the first time you ever saw an item, and it is the number closest to unseen ability — but only within the topics you have attempted, and only if you answer before reading the explanation. Repeat accuracy is your score on items you have already seen; it climbs quickly because you are recognising a specific stem, so a near-ceiling score means little. A wide gap between the two says you are memorising items, not mastering clinical reasoning and appraisal.
Peer histogram and percentile. The daily-updated histogram plots you against other subscribers who chose the same items, not the national Paper B cohort, so it is directional rather than absolute and shifts with who else is revising. There is no formal national percentile and no published predicted pass score; do not read the histogram as a probability of passing, and treat any predicted score as motivational rather than diagnostic.
Coverage is the share of the blueprint you have attempted at all — and for Paper B the coverage question that matters most is whether the critical-review third is proportionally represented. Difficulty is something you largely control through topic selection, so your running percentage partly reflects the mixture you chose. Time per item matters because the real paper gives you 150 questions in 180 minutes — about 72 seconds each — and statistics items with a data table, plus EMIs, can eat that budget quickly; only timed mode shows whether you are inside it.
Why the headline percentage is not your exam score
Two mechanisms bias the pool the dashboard scores. First, you choose the topics, and most candidates gravitate to the clinical psychiatry they enjoy and away from statistics and critical appraisal, so a rising percentage can simply mean you have stopped attempting a third of the exam. Second, the strength-and-weakness prompts, used naively, encourage isolated grinding of weak areas — useful for learning, but it makes your headline number non-representative. Either way, a percentage on a self-selected pool is not comparable with an unseen, blueprint-mixed, timed block — the argument in "Your Q-Bank Percentage Is Not Your Exam Score". For Paper B the denominator problem has a face: the candidate whose 75% hides that they have barely touched the critical-review items.
The blueprint audit MRCPsychMentor will not run for you
Instead of trusting the home-screen average, lay your attempted-question distribution beside the official Paper B weighting. Here, unusually, one weight is known and fixed: critical review is about a third of the paper.
| Paper B area | Official weight | Example attempted share | Flag |
|---|---|---|---|
| Critical review, research methods and statistics | ~1/3 (≈50 of 150) | 17% | Under-sampled — the classic gap |
| General adult and old age psychiatry | Verify (RCPsych blueprint) | 41% | Over-sampled |
| Child and adolescent, forensic, learning disability | Verify (RCPsych blueprint) | 24% | Check accuracy |
| Substance misuse, psychotherapy, service organisation | Verify (RCPsych blueprint) | 18% | On target |
Fill this in from your own account data. The classic Paper B failure is a candidate strong across general adult and old age who has attempted only a fraction of the statistics and appraisal items — then forfeits a third of the marks the exam guarantees. The method is the blueprint-coverage matrix from "Question-bank completion is not coverage"; the bank gives you the attempts, you build the matrix.
The readiness test — five conditions for a credible signal
A number only counts as a readiness signal when it is earned under exam-like conditions. Require all five: unseen (never-attempted items, not flagged re-runs); timed at roughly 72 seconds per item; mixed across clinical and critical-review content; sat with no assistance (no explanations, no paused clock); and a sufficiently large sample (a full 150-item mock or several mixed blocks). A 78% on a topic-filtered, untimed clinical quiz is not the same evidence as 78% on a fresh, timed, mixed, full-length block that includes its proper share of statistics.
Algorithm override rules
Because you drive the topic selection, the most testable material can slip through. Set a weekly quota and force these in: the whole critical-review third — study design, bias and confounding, sensitivity, specificity and predictive values, number needed to treat, meta-analysis and forest plots, and the arithmetic behind them; the EMI format itself; low-volume clinical corners such as learning disability, perinatal and neuropsychiatry; and service-organisation and ethics content dismissed as "not real psychiatry" but which carries reliable marks. A rising overall percentage must never hide the critical-review items you have never attempted.
Worked dashboard example: from analytics to next week's quotas
Picture a candidate five weeks out: overall first-attempt 70%; general adult 78%; old age 72%; child and adolescent 62%; forensic 64%; critical review and statistics 54% on only about 15% of attempts; time per item 66 seconds untimed. Read it honestly — the critical-review score is both low and thinly sampled, so the 70% is propped up by comfortable domains, and the untimed pace hides how a data-table item behaves under the clock. Next week's quotas then write themselves, with no pass prediction attached.
| Domain | Dashboard reading | Next-week quota | Mode |
|---|---|---|---|
| Critical review and statistics | 54% first-attempt, ~15% of attempts | 120 new items | Topic-filtered, then timed |
| Child and adolescent, forensic | 62–64% first-attempt | 60 new items | Topic-filtered |
| General adult | 78% first-attempt | 30 items | Timed mixed |
| Whole-blueprint check | not measured cold | one 150-item timed mixed mock | Unseen, no filters |
Nothing in that plan claims the candidate will pass; it simply moves effort to the measured gap — here, the guaranteed critical-review third — and forces one honest, unseen, timed check.
Seven-day pattern: MRCPsychMentor for one job, iatroX for unseen transfer
This is for a trainee revising around clinical work, using MRCPsychMentor for one job — building and repairing written recall — and iatroX for unseen transfer measurement. No proprietary-algorithm claims are made for either tool.
- Day 1 (Mon) — MRCPsychMentor: 40 critical-review and statistics items, topic-filtered, every miss reviewed and logged as a one-line rule (for example, the difference between relative and absolute risk).
- Day 2 (Tue) — 40 general-adult and old-age items; tag each error by type.
- Day 3 (Wed) — 30 child-and-adolescent and forensic items plus a spaced re-test of Monday's statistics misses; then switch tools for a 30-item unseen, timed, mixed iatroX Paper B block and record first-attempt accuracy by domain.
- Day 4 (Thu) — 40 more critical-review items on the weakest sub-topics — study design, diagnostic-test statistics — redoing flagged items.
- Day 5 (Fri) — one timed MRCPsychMentor mock under full conditions.
- Day 6 (Sat) — review the mock; two short source reads on the worst topics, taking any medicines detail from the SmPC or electronic medicines compendium rather than memory.
- Day 7 (Sun) — a 60–90-item unseen, timed, mixed iatroX block; compare with Wednesday to see whether the week transferred, watching the critical-review sub-score in particular.
That is the two-bank pattern from "The two-Q-bank rule": one bank to learn on, a different, unseen bank to measure on. The companion MRCPsychMentor Paper B adaptive workflow gives the fuller step-by-step.
Decision checklist: continue, supplement, switch or stop
| Signal (measurable, not sunk cost) | Action |
|---|---|
| First-attempt accuracy rising across clinical and critical-review content; critical-review items attempted in proportion; unattempted first-pass items remain | Continue |
| Bank percentage high but unseen mixed first-attempt lags by more than 10 points; critical review under-sampled | Supplement — add unseen mixed blocks and force the statistics third |
| Coverage floor met and first-attempt stable, but pace runs over 72 seconds or accuracy falls under time on data items | Switch emphasis to timed mixed mocks |
| Repeat accuracy near the ceiling on almost everything, first-attempt plateaued, unseen blocks confirm the standard | Stop drilling; taper and rest |
Bottom line
MRCPsychMentor is a large, syllabus-mapped Paper B bank with a useful peer-comparison histogram, category-weighted mocks and per-question explanations, and a strong option for the written paper. But its dashboard measures your training pool, not the exam, and it does not, on its public pages, describe a proprietary adaptive-difficulty engine or a predicted score. Use it to find and drill weak, thin domains — above all the critical-review third the exam guarantees — and use a separate unseen, mixed, timed source, alongside the College's materials, to decide whether you are ready; it does nothing for the CASC. The number that forecasts your exam is the one you earn on questions the system did not choose for you.
Frequently asked questions
Is MRCPsychMentor enough for MRCPsych Paper B on its own? For the written paper, over 2,000 MCQ and EMI items (vendor-reported, 20 July 2026) can serve many candidates as a main bank — provided they genuinely work the critical-review third rather than avoiding it. But "enough" turns on two things it does not do: it does not measure you on unseen, timed, blueprint-mixed items once you start recognising questions, and it does nothing for the CASC. Treat it as a strong single bank for one component, and verify the current count on the product page.
Which MRCPsych Paper B component does MRCPsychMentor not reproduce well? Two things. The clinical CASC, which tests spoken assessment and communication no MCQ rehearses — the vendor's separate AI CASC tool needs its automated marking calibrated against a human before you trust it, as we explain in calibrating AI-graded feedback. And within the written paper, the end-to-end reasoning of critical appraisal: a recognition item can test whether you know what a confidence interval is, but it under-rehearses appraising an unfamiliar paper in full, which is closer to what the exam demands.
How many MRCPsychMentor questions should I complete per day for MRCPsych Paper B? There is no magic number; first-attempt quality and whether the critical-review third is represented matter more than volume. A sustainable pattern around clinical work is roughly 30–50 new first-attempt items on weekdays with full review of every miss, rising at weekends with a timed mock, and a standing rule that at least a third of your weekly items are statistics and appraisal. Forty analysed questions beat 150 rushed ones you never revisit.
When should I stop using MRCPsychMentor and move to mixed mocks? Move when three measurable conditions hold: you have attempted at least about 80% of each clinical area and of the critical-review content, your first-attempt accuracy has stabilised, and you have begun recognising items rather than reasoning them out. At that point repeat accuracy climbs for the wrong reason, so shift the centre of gravity to timed, random, full-length mixed mocks — the College's materials and a second unseen bank — to test transfer, pacing and that the statistics third holds up cold.
How should I combine MRCPsychMentor with iatroX without duplicating practice? Give each tool one job. Use MRCPsychMentor as your build-and-repair bank — first-pass learning, explanations and spaced review of misses, including the critical-review items — and iatroX only as the measurement bank: fresh, unseen, timed, mixed blocks run once or twice a week. Because the item pools differ you are not re-answering the same questions; you are separating "have I learned it" from "can I retrieve it cold", the point of the two-Q-bank rule.
Editorial notes and references
Written by Dr Kolawole Tytler, NHS GP and founder of iatroX. Last checked 20 July 2026; MRCPsychMentor question counts, prices and features are vendor-reported as of that date and change without notice — verify the current figures on the product page before purchase. Disclosure: iatroX operates a competing MRCPsych Paper B question bank, so this audit confines iatroX's role to the unseen-measurement job MRCPsychMentor does not claim to perform, and takes no position on price. Corrections are welcome via the feedback route on iatrox.com.
References: Royal College of Psychiatrists — MRCPsych examinations, syllabus and sample papers (rcpsych.ac.uk; verify the current format, the critical-review weighting and standard-setting); MRCPsychMentor Paper B product page (mrcpsychmentor.com; vendor-reported figures); iatroX MRCPsych Paper B bank (https://www.iatrox.com/mrcpsych-paper-b); iatroX comparison hub (https://www.iatrox.com/compare); "Your Q-Bank Percentage Is Not Your Exam Score" (https://www.iatrox.com/blog/qbank-percentage-not-your-exam-score).