This workflow is for core psychiatry trainees using BMJ OnExamination for MRCPsych Paper A — the written paper on the scientific and theoretical basis of psychiatry. It addresses the MCQ and extended-matching written paper only. The principal limitation to state first: BMJ OnExamination has no proprietary adaptive engine, so the "adaptive" part of this plan is something you drive by hand against the blueprint, and — like any single bank — its value collapses if you burn through your only unseen questions early and are left re-testing on items you already recognise.
What BMJ OnExamination offers for MRCPsych Paper A right now
As of 19 July 2026, BMJ OnExamination lists an MRCPsych Paper A bank of 970+ questions (vendor-reported), mapped to the current curriculum. Subscriptions run from about £49.99 for one month to £149.99 for twelve months (vendor-reported), with ten free questions a day on trial. Features include mock tests built around recent exam themes, a "Select Questions" tool to choose difficulty and recap items you found hard, detailed explanations with a Key Learning Point, a group-learning leaderboard, and an offline mobile app. Verify the current count, price and features on onexamination.com before subscribing.
An honest note on "adaptive": the product page describes user-directed personalisation — you pick difficulty and recap flagged questions — not an automated adaptive-difficulty algorithm. The plan below therefore keeps you in charge of coverage rather than delegating it, which is exactly what protects the low-frequency domains.
MRCPsych Paper A: the official shape of the exam
Paper A is a written paper of multiple-choice questions and extended matching items — roughly two-thirds MCQ and one-third EMI — of about 150 questions over three hours, delivered by computer at Pearson VUE. It covers the scientific and theoretical basis of psychiatry: behavioural science and sociocultural psychiatry, human development, basic and clinical neurosciences and neuroanatomy, genetics, psychopharmacology, classification and assessment, and the basic psychological processes and theories. Confirm the current question count and duration on rcpsych.ac.uk. The clinical CASC (an OSCE-style examination) sits separately and later in the membership pathway; this workflow does not address it, and iatroX is not a CASC simulator — it is a written-knowledge and unseen-MCQ layer.
For any psychopharmacology fact, the UK medicines reference is the SmPC via the electronic medicines compendium (eMC), read with NICE and SIGN guidance — not a formulary shorthand.
Baseline week: measure before you personalise
Before you let the interface, difficulty selector or leaderboard shape your feed, complete a small blueprint-stratified unseen sample — around 60–90 questions deliberately spread across the Paper A domains, with fixed small blocks for the areas trainees most often defer: neuroanatomy and neurophysiology, genetics, statistics-adjacent research methods, and the psychological theories. Record first-attempt accuracy per domain. That baseline is what a rising overall percentage will otherwise conceal.
First pass: set domain floors
The characteristic Paper A trap is that psychopharmacology and clinical-adjacent items feel familiar, so you answer more of them, the overall score rises, and neuroscience or developmental theory stays at zero attempts. Prevent it with domain floors: a minimum number of attempted questions in every blueprint area — say no domain below 40 items — before the headline number counts for anything. Use "Select Questions" to service the floors deliberately, not to linger in comfortable topics.
Error taxonomy: label every miss
Sort each wrong answer into one of six causes, because the remedy differs:
- Knowledge gap — the fact, mechanism or theory was not known.
- Misread stem — the vignette or lead-in was misread, especially in EMIs with long option lists.
- Premature closure — you fixed on an answer before reading the whole item.
- Guideline/classification error — an outdated criterion or a NICE/ICD/DSM mismatch.
- Calculation error — a genetics-, epidemiology- or pharmacokinetics-style numeric slip.
- Time-pressure error — a mistake that only appears under the clock.
Tally causes weekly. A pile of knowledge gaps needs reading; a pile of misreads and premature closure needs timed practice and EMI technique, not more content.
Review interval: match the fix to the error
Resist the urge to repeat the same item immediately — that mostly trains recognition of that item. A knowledge gap earns a short source read plus a new transfer question on the concept; a classification or guideline error earns a primary-source check and a spaced review in a few days; a misread or premature-closure error earns timed mixed practice; a calculation slip earns a drilled worked method. Keep immediate repeats for genuine one-off slips.
Mixed-block switch: when to stop filtering
Filtered practice builds; mixed random blocks test. Switch once all domain floors are met and per-domain first-attempt accuracy sits within a narrow band of target. The objective trigger: when no single domain is dragging and your errors are increasingly misreads and pacing rather than knowledge, cut filtered practice and increase full-length, timed, random blocks that mirror the ~150-question paper.
Exit criteria: what "ready" actually means
Bank completion is not readiness. Exit when five things hold: every blueprint domain clears its coverage floor; first-attempt accuracy on unseen items is stable across two or more sittings; you are pacing at roughly a question a minute with time to spare; retention holds on spaced re-tests, not fresh reads; and you have calibrated against the College's own sample questions, not just the commercial bank. One Paper-A-specific caution: basic-science recall — neuroanatomical tracts, receptor pharmacology, developmental milestones — decays faster than clinical pattern recognition, so lean on spaced re-tests rather than a single confident pass, and treat a domain as truly cleared only when it survives a delayed retest a week or so later.
Worked example: a seven-day plan
A core trainee, six weeks out, revising around clinical work. BMJ OnExamination does one job — MCQ/EMI volume and topic drill — and iatroX supplies unseen transfer measurement. No proprietary-algorithm claims. The point of the split is that a bank you learn on cannot also be the clean instrument that tells you whether the learning stuck.
- Day 1: Blueprint-stratified baseline (about 75 items); log per-domain first-attempt accuracy.
- Day 2: Service the two weakest basic-science domains to their floors; error-tag every miss.
- Day 3: Source reads for day-2 knowledge gaps; check any drug fact against the SmPC/eMC.
- Day 4: A timed mixed block; separate misreads and premature closure from true gaps.
- Day 5: A fresh, unseen, timed MRCPsych Paper A block in iatroX — no filtering, no assistance — to measure transfer.
- Day 6: Spaced review of the week's classification and calculation errors; a short EMI technique drill.
- Day 7: Recompute domain floors and the error tally; set next week's minimums.
Decision checklist: continue, supplement, switch or stop
- Continue if floors are filling and unseen first-attempt accuracy is trending up.
- Supplement with a second, unseen bank if your BMJ score is high but transfer to fresh items lags, or a basic-science domain is thin.
- Switch emphasis to mixed timed practice once floors are met and knowledge gaps are shrinking.
- Stop filtered practice in the final fortnight; move to timed, mixed, unseen blocks plus official-material calibration.
Decide on measurable gaps, not on how many questions remain unticked.
Bottom line
BMJ OnExamination is a large, curriculum-mapped bank and a reasonable spine for Paper A's basic-science load — as long as you drive it by the blueprint, keep domain floors so a rising average cannot bury neuroscience or developmental theory, and preserve a clean unseen set to measure transfer. Learn on the bank; measure elsewhere; and remember Paper A is only the written knowledge stage, with the CASC to prepare separately.
FAQ
Is BMJ OnExamination enough for MRCPsych Paper A on its own? For the written knowledge it can be a strong spine, but "enough on its own" overstates it: any single bank drifts toward recognition as you finish it, and you still need the College's own sample questions for calibration and a clean unseen set to confirm transfer. It also does not touch the CASC. Used as one component with those additions, it is a solid option.
Which MRCPsych Paper A component does BMJ OnExamination not reproduce well? It reproduces the MCQ and EMI written format well, but it cannot prepare you for the CASC — the clinical OSCE-style examination is a different skill entirely — and no MCQ bank should be treated as CASC practice. Within Paper A, watch that basic-science domains such as neuroanatomy and genetics are genuinely covered rather than thinly represented behind the headline total.
How many BMJ OnExamination questions should I complete per day for MRCPsych Paper A? There is no official number; reviewed quality beats volume. Many trainees manage roughly 40–60 questions a day with full review and error-tagging, which teaches far more than a larger unreviewed block. Prioritise clearing domain floors and verifying misses over a daily count, and protect at least one timed mixed block each week.
When should I stop using BMJ OnExamination and move to mixed mocks? Once every blueprint domain clears its floor and unseen first-attempt accuracy is stable, filtered single-topic practice yields little. In the final two to three weeks the binding constraints are pacing across roughly 150 questions in three hours and retention, so shift to timed, mixed, unseen blocks and the College's own material for calibration.
How should I combine BMJ OnExamination with iatroX without duplicating practice? Give each bank one job: learn and drill on BMJ OnExamination, and reserve iatroX for fresh, unseen, timed blocks that measure transfer — never re-enter a BMJ item into iatroX or the reverse. That separation, set out in the iatroX two-Q-bank rule, keeps calibration honest: one bank to build on, one unseen bank to measure on.
Editorial notes and references
Written by Dr Kolawole Tytler, NHS GP and founder of iatroX. Last checked 19 July 2026; product figures are vendor-reported and change without notice, so verify the question count, price and features on onexamination.com before purchase, and confirm the exam format on rcpsych.ac.uk. Disclosure: iatroX operates a competing MRCPsych question bank; this article confines iatroX to the unseen-measurement role BMJ OnExamination does not claim to fill, is explicit that iatroX is not a CASC simulator, and makes no proprietary-algorithm claims. Corrections are welcome via the feedback route on iatrox.com. References: Royal College of Psychiatrists MRCPsych Paper A syllabus and sample questions (rcpsych.ac.uk); NICE, SIGN and the electronic medicines compendium (medicines.org.uk) for the SmPC; the iatroX pillar on building a blueprint-coverage matrix and "Your Q-Bank Percentage Is Not Your Exam Score"; iatroX MRCPsych Paper A bank and comparison hub.