This is an implementation guide, not a review — if you want the analytics audit, read what PassMedicine's engine is actually optimising first. Here the job is narrower: a concrete PassMedicine workflow for MRCP Part 1 that tells you when to follow your practice loop, when to override it, and when to stop drilling and start simulating. The principal limitation it works around is that a self-directed bank rewards comfort, and comfort is not the blueprint.
What you are working with
As of 19 July 2026, PassMedicine's MRCP Part 1 bank offers 5,100+ best-of-five questions, two textbooks, three timed mocks and peer-comparison analytics, at £35 for four months or £45 for six (verify before purchase). No AI tutor or ML adaptive engine is advertised; you configure the loop through filters, modes and repeats. The exam you are configuring for is two three-hour papers of 100 best-of-five questions, with clinical sciences (25) and pharmacology (15) dominating the per-paper blueprint and the organ specialties trailing down to ophthalmology and palliative medicine at 4 each.
Week structure at a glance
| Phase | Weeks (typical 10-week run) | Primary mode | Exit trigger |
|---|---|---|---|
| Baseline | 1 | Unseen, blueprint-stratified sample | ~100 questions logged across domains |
| First pass | 2–6 | Domain-floored revision + explanations | Every domain above coverage floor |
| Consolidation | 7–8 | Timed topic blocks + error transfer | First-attempt accuracy stable 2 weeks |
| Simulation | 9–10 | Full mixed timed mocks | Pacing on target; go/no-go |
Adjust the calendar to your runway, but keep the order: you cannot judge a personalised feed you never baselined, and you cannot simulate an exam you have not first covered.
Baseline before you let it personalise
Spend the first week generating a small, blueprint-stratified unseen sample — a fixed number of questions per specialty in rough proportion to the exam, answered untimed but honestly, before you lean on any filtered revision. This gives you a true starting distribution of strengths and weaknesses rather than one shaped by which topics you reached for first. Without a baseline, every later percentage is contaminated by selection, and you will mistake "the topics I chose" for "the topics I know".
First pass: set domain floors
The single most important override: set a minimum question count per blueprint domain and do not let the overall percentage rise until every floor is met. Left alone, a self-directed loop lets an improving score coexist with unattempted syllabus areas — you feel progress while blind spots survive. Floors force breadth. Set them proportional to the blueprint (higher for clinical sciences and pharmacology, non-zero for ophthalmology, oncology and palliative care) and treat an unmet floor as a hard block on declaring readiness, whatever the dashboard average says.
An error taxonomy that dictates the fix
Not every wrong answer means "revise this topic". Sort each miss into one of six types, because each demands a different response. A knowledge gap needs a source read and a spaced repeat. A misread stem needs a process fix (read the last line first), not more content. Premature closure needs deliberate differential-listing practice. A guideline error needs the current source, checked and dated. A calculation error needs drilled arithmetic under time. A time-pressure error needs pacing work, not knowledge work. Logging misses by type turns a vague "I got 60%" into a specific action list — and stops you re-revising things you actually know but keep misreading.
Review interval: match the response to the error
Decide, per error, whether it earns an immediate repeat (rarely — that trains recognition), a new transfer question on the same concept (usually the best choice), a scheduled spaced review, or a short source read. The default should be transfer, not repeat: seeing the same item again inflates your repeat-accuracy metric while teaching you the item rather than the concept. A fresh question on the same idea — ideally from an unseen pool — is what proves the gap is closed.
The mixed-block switch: objective criteria
Move from topic-filtered practice to timed random blocks when three conditions hold: every domain floor is met, first-attempt accuracy on unseen material has been stable for two weeks, and your per-item pace is within exam budget (~1.8 minutes). Not before — mixed blocks are a measurement instrument, and measuring before you have covered the syllabus just produces a discouraging number with no diagnostic value. Once the criteria are met, invert the ratio: mostly mixed timed blocks, topic practice only to patch specific leaks the blocks reveal.
Exit criteria: what "ready" actually means
Ready is not "finished the bank" and not "hit X%". It is five things together: coverage floor met in every domain; first-attempt accuracy stable across two weeks of unseen mixed blocks; pacing on target; retention holding on spaced reviews of earlier errors; and a calibration check against official MRCP sample material (not just third-party questions). Miss any one and you have a specific job left; hit all five and further grinding has low marginal value.
A seven-day pattern for busy trainees
Monday: 40 PassMedicine questions in two floor-deficient domains, explanations read, misses taxonomised. Tuesday: 30 questions plus transfer questions for Monday's knowledge-gap errors. Wednesday: a timed, unseen 50-question mixed block in iatroX's free MRCP Part 1 bank — your uncontaminated readiness signal and a source of transfer questions PassMedicine repeats cannot provide. Thursday: light spaced review of the week's errors only. Friday: 40 PassMedicine questions on the next floor-deficient domains, timed. Saturday: a PassMedicine mock or second mixed block; same-day review by error type. Sunday: rest. PassMedicine supplies volume, floors and explanations; iatroX supplies unseen transfer and measurement; your error taxonomy is the steering wheel.
Three failure patterns this workflow prevents
Most MRCP Part 1 near-misses trace to one of three habits, and each maps to a specific step above. The first is coverage drift: a candidate rides an improving overall percentage while clinical sciences and pharmacology — 40 of the exam's 200 marks — stay under-attempted, because those domains are drier than the organ specialties and the feed never forces them. Domain floors exist precisely to make this impossible; if you set them proportional to the blueprint and refuse to declare readiness until they are met, coverage drift cannot survive. The second is repeat inflation: a candidate re-answers previously-seen items, watches repeat accuracy climb into the high 80s, and mistakes recognition for knowledge. The transfer-question rule kills this — a fresh question on the same concept, ideally unseen, is the only thing that proves a gap is closed. The third is the pace ambush: a candidate with excellent untimed accuracy meets the real paper's 1.8-minutes-per-item budget for the first time on exam day and finishes two questions short in each paper. The mixed-block switch, run at full pace, surfaces this weeks early when it is still fixable.
Notice that none of these failures shows up as a bad dashboard number. All three are compatible with a reassuring overall percentage, which is exactly why the workflow anchors on floors, transfer and timed blocks rather than on the headline figure the platform puts in front of you.
Continue, supplement, switch or stop
Continue while floors are filling and unseen performance climbs. Supplement with unseen transfer blocks whenever repeat accuracy outpaces first-attempt accuracy. Switch only for a measurable gap, never for novelty. Stop drilling and commit to simulation once all five exit criteria hold — usually the final fortnight.
Frequently asked questions
Is PassMedicine enough for MRCP Part 1 on its own? It can carry the bulk of question volume, but the workflow above still needs unseen transfer questions and official sample material to close the loop, because a single bank cannot both drill you and independently measure you.
Which MRCP Part 1 component does PassMedicine not reproduce well? The blueprint's enforced breadth under unseen timed conditions — which is exactly why domain floors and a hard mixed-block switch are built into this workflow.
How many PassMedicine questions should I complete per day for MRCP Part 1? 40–60 on weekdays and 80–100 on a weekend day is a sustainable frame; distribute by floor deficits and protect two timed mixed blocks weekly regardless.
When should I stop using PassMedicine and move to mixed mocks? When domain floors are met, first-attempt accuracy is stable for two weeks and pacing is on target — invert to mostly-mixed at that point, typically the last two to three weeks.
How should I combine PassMedicine with iatroX without duplicating practice? PassMedicine for floored domain drilling and explanations; iatroX for unseen transfer questions, adaptive coverage of related weaknesses and timed measurement — no question gets practised twice.
Editorial notes and references
Written by Dr Kolawole Tytler, NHS GP and founder of iatroX. Last checked 19 July 2026; PassMedicine figures are vendor-published — verify before purchase. Disclosure: iatroX operates a free competing MRCP Part 1 bank. Corrections via the feedback route on iatrox.com. References: MRCP(UK) Part 1 format and blueprint (thefederation.uk); PassMedicine MRCP page (passmedicine.com/mrcp); related reading: the PassMedicine MRCP analytics audit and why your Q-bank percentage is not your exam score.