PassMedicine is a strong, low-cost core bank for MRCP Part 1 candidates — but the numbers on its dashboard measure practice performance, not exam readiness, and the two diverge in predictable ways. This audit is for candidates already using or about to buy PassMedicine who want to know what its analytics and question-selection features actually optimise, and the principal limitation is simple: a rising overall percentage can coexist with untouched blueprint domains.
What PassMedicine offers for MRCP Part 1 right now
Checked against the vendor's public product page on 19 July 2026: over 5,100 best-of-five questions, sold as fixed-term access at £35 for four months or £45 for six. The package includes two textbooks (a syllabus-focused and an extended version), three mock exams built to the real exam's timing and question distribution, revision and timed test modes, and performance analytics with peer-comparison histograms. PassMedicine states its questions are "constantly reviewed" against current guidelines.
Two honest observations before the audit. First, PassMedicine does not advertise an AI tutor or a machine-learning adaptive engine for MRCP Part 1 — the "engine" in this article's title is the practice loop you configure yourself through its filters, repeat options and analytics, and it is worth being precise about that because candidates routinely attribute more intelligence to the selection process than the platform claims. Second, at this price the value is unarguable; nothing in this audit is an argument against buying it. It is an argument against misreading it.
The exam the numbers must answer to
MRCP(UK) Part 1 is two three-hour papers, 100 best-of-five questions each, no negative marking. The Federation publishes the per-paper specialty distribution, and it should anchor every revision decision: clinical sciences 25 questions, clinical pharmacology and therapeutics 15, then cardiology, infectious diseases, neurology, renal, respiratory, rheumatology, gastroenterology and endocrinology at 14 each, haematology 10, psychiatry 9, dermatology and geriatric medicine 8 each, oncology 5, medical ophthalmology 4 and palliative medicine 4.
Notice what that table implies: clinical sciences plus pharmacology are nearly a fifth of the paper — more than any single organ specialty — and the small domains (ophthalmology, palliative care, oncology) still contribute 13 marks across both papers. Candidates fail margins smaller than that.
What each dashboard metric actually measures
First-attempt accuracy is the only number on the dashboard that behaves like the exam, because the exam is entirely first attempts. Repeat accuracy measures recognition of previously seen items — it rises with exposure even when underlying knowledge has not moved, which is why overall percentage drifts upward through a subscription almost regardless of learning. Peer-comparison histograms tell you where you sit relative to other PassMedicine users answering the same items — a useful sanity check, but the comparison population is self-selected and includes people at every stage of preparation, so a comfortable percentile is not a pass probability. Coverage counts questions attempted per specialty, which is an input measure, not an outcome. Difficulty and time-per-item are diagnostic gold when read per domain: accurate-but-slow in a specialty is a different problem from fast-but-wrong, and the two need different fixes.
The metric PassMedicine does not show you — because no bank can — is transfer: whether you can answer an unseen question on the same concept phrased a different way. That is the property the exam tests, and it is the property a separate, unseen question source measures best.
Selection bias: why your percentage flatters you
Any practice loop that lets you choose what to answer produces biased statistics, whatever the platform. Answer mostly your strong specialties and your average inflates; drill your weak ones (as diligent candidates do) and your average deflates just as misleadingly, because you are comparing a weak-domain-heavy sample against other people's mixed samples. Repeat previously wrong questions — an excellent learning habit — and your repeat accuracy climbs while telling you nothing about the exam. The general rule: any percentage produced under conditions you chose is not comparable with a mixed, unseen, timed block, and only the latter resembles the exam. This is the core reason we wrote Your Q-Bank Percentage Is Not Your Exam Score; it applies to every bank we audit, PassMedicine included.
Audit your distribution against the blueprint
Once a fortnight, pull your per-specialty attempted counts and set them against the official distribution above. The calculation takes five minutes: each specialty's share of your total attempts, versus its share of the exam's 200 questions. Candidates who do this usually find the same pattern — organ specialties they enjoy sit at twice their exam weight, while clinical sciences, pharmacology, ophthalmology and palliative care sit under-attempted at a fraction of theirs. The home-screen average is silent about this because it weights by what you did, not by what the exam will do.
What a credible readiness signal requires
Five conditions, all of them: unseen questions, timed at exam pace (roughly 1.8 minutes per item), mixed across specialties in blueprint-like proportions, no mid-block look-ups, and a sample large enough to mean something — a single 30-question block has a confidence interval wide enough to swallow a grade boundary, so judge on 100+ questions across multiple sittings. PassMedicine's three mocks satisfy these conditions and are the most exam-like signal inside the product; its filtered revision modes, by design, satisfy none of them. Both are useful. Only one is evidence.
When to override the selector
Configure your own practice against the blueprint rather than your instincts. Force clinical sciences and pharmacology blocks weekly regardless of comfort, because 40 marks live there. Force the micro-domains — ophthalmology, palliative medicine, oncology — on a rota, because low-volume domains generate few questions in any feed and are easiest to leave untouched. Force image-based and calculation items if your attempted mix under-represents them. And periodically force a full random block precisely because it ignores your preferences: it is the closest thing a bank offers to the exam's indifference to what you feel like revising.
A worked dashboard example
Suppose after four weeks your dashboard shows: overall 68%, cardiology 78% (412 attempts), gastroenterology 74% (301), clinical sciences 55% (88), pharmacology 58% (102), ophthalmology 62% (13), palliative care no attempts, average 2.4 minutes per item in neurology with 71% accuracy.
Read properly, that is not "68% — on track". It is: cardiology is over-practised and can be maintained with one small weekly block; clinical sciences and pharmacology are under-sampled and under-performing on nearly 20% of the exam, so they take the largest quota next week (say 120 questions between them); ophthalmology's 62% is thirteen questions of noise, not a measurement — assign a block; palliative care is a blank spot — assign its first; neurology's problem is pace, not knowledge, so it gets timed blocks specifically. Next week's plan writes itself as quotas, and no pass probability was needed at any point.
A seven-day pattern for busy trainees
A realistic on-call-compatible week, using each tool for one defined job: Monday, 40 PassMedicine questions in your two forced weak domains, untimed, reading every explanation. Tuesday, 30 more plus a 15-minute error-list review. Wednesday, a timed 50-question mixed block in iatroX's free MRCP Part 1 bank — unseen items, adaptive selection that probes related weaknesses across topic boundaries, and a first-attempt signal PassMedicine repeats can no longer give you. Thursday, rest or 20 light questions. Friday, 40 PassMedicine questions in the blueprint domains your audit flagged. Saturday, a PassMedicine mock or a second timed mixed block, reviewed the same day by error type. Sunday, spaced review of the week's errors only. PassMedicine does volume and explanations; iatroX does unseen transfer testing and adaptive coverage; neither is asked to do the other's job, and nothing here requires claims about anyone's proprietary algorithm.
Continue, supplement, switch or stop
Continue if unseen mixed performance is rising and your blueprint audit is roughly level. Supplement (with unseen mixed blocks and transfer practice) if the bank's numbers rise while unseen performance stalls — that is recognition, not learning. Switch only for a measurable gap: a domain the bank covers thinly for you, or analytics you cannot get; novelty is not a reason and sunk cost is not a reason to stay. Stop question-grinding and move to mock conditions in the final fortnight when coverage is complete, first-attempt accuracy is stable and pacing is on target — completion of a bank is not an exit criterion, and neither is a round number on a dashboard.
Frequently asked questions
Is PassMedicine enough for MRCP Part 1 on its own? For many candidates it can carry the bulk of preparation given its volume, price and explanations, but it cannot provide the one measurement that predicts the exam — performance on unseen, mixed, timed material — so at minimum add independent mock-condition practice and treat its dashboard as practice data rather than a readiness verdict.
Which MRCP Part 1 component does PassMedicine not reproduce well? The exam's indifference to your preferences: its fixed blueprint weighting of clinical sciences, pharmacology and the small domains, delivered as unseen items under time pressure. Filtered revision inside any bank, PassMedicine included, systematically under-rehearses exactly that.
How many PassMedicine questions should I complete per day for MRCP Part 1? Volume matters less than composition, but as a working figure 40–60 on weekday study days and 80–100 on a weekend day is sustainable alongside clinical work; distribute them by your blueprint audit, not by comfort, and protect two timed mixed blocks per week whatever the daily count.
When should I stop using PassMedicine and move to mixed mocks? When three things hold simultaneously — every blueprint domain attempted above your coverage floor, first-attempt accuracy stable across two consecutive weeks, and pacing at or under 1.8 minutes per item — shift the balance to timed mocks and unseen blocks, keeping the bank only for reviewing errors. For most candidates that is the final three to four weeks.
How should I combine PassMedicine with iatroX without duplicating practice? Give each a non-overlapping job: PassMedicine for high-volume domain drilling and its explanations; iatroX for unseen timed blocks, adaptive selection that surfaces related weaknesses you did not choose, and free MRCP Part 1 coverage — so your readiness signal always comes from questions neither you nor your history selected.
Editorial notes and references
Written by Dr Kolawole Tytler, NHS GP and founder of iatroX. Last checked 19 July 2026; PassMedicine figures above are vendor-published and were checked on that date — confirm current prices and counts on the product page before purchase. Disclosure: iatroX operates a question bank covering the same exam, which is why this audit names PassMedicine's strengths plainly and confines the iatroX role to jobs the audited product does not claim to do. Corrections are welcome via the feedback route on iatrox.com. References: MRCP(UK) Part 1 format and blueprint, thefederation.uk/examinations/part-1/format; PassMedicine MRCP product page, passmedicine.com/mrcp; our related pieces on the best MRCP Part 1 banks and whether PassMedicine is worth it in 2026, and the comparison hub.