This audit is for candidates deciding whether to rely on MRCEM Exam Prep as their primary MRCEM SBA (Intermediate) bank. In short: it is a high-volume, low-cost written-question resource that is a reasonable first-pass and breadth tool, and its principal limitation is the one every static bank shares — it measures recall and single-item reasoning, not the timed, mixed, whole-paper prioritisation that the two 90-item papers actually test. Treat it as breadth, and measure transfer elsewhere.
Current state (last checked 19 July 2026)
| Attribute | Reported position |
|---|---|
| Exam components | MRCEM Primary and MRCEM SBA (Intermediate) — the SBA product is the relevant one here |
| Live question count | 2,600+ single-best-answer items for the SBA (vendor-reported) |
| Written tutorials | 310+ revision tutorials aligned to the clinical syllabus (vendor-reported) |
| Price | From £35 for the SBA course (vendor-reported; depends on access window) |
| Access period | Not clearly stated on the pages checked — verify on the product page |
| AI / adaptive features | None advertised on the pages checked; treat as a static bank |
| Free trial | Free starter pack reported |
All figures are vendor-reported and change; confirm the count, price and access length on mrcemexamprep.net before purchase.
Exam anchor: what the SBA actually requires
The SBA is 180 single-best-answer questions across two two-hour papers of 90, roughly 80 seconds per item, mapped to the 2021 RCEM curriculum (Year 1–3 SLOs). The domain weighting — Complex/Stable 55, Resuscitation 40, Injured 30, Paediatric EM 25, Procedural 20, Complex/Challenging 10 — is the yardstick. There is no negative marking, the pass mark is set by modified Angoff, and MRCEM Primary is a prerequisite. RCEM publishes sample SBAs and a regulations pack, and that official material — not any vendor claim — defines the question style and difficulty you are calibrating to. Treat any "mapped to the curriculum" statement from a third party as a claim to verify, not a guarantee.
Reading the headline count against the blueprint
A 2,600-item total tells you little on its own; what matters is distribution against the weighting. The bank exposes domain filters — use them to build the six-domain table (Complex/Stable 55, Resus 40, Injured 30, Paeds EM 25, Procedural 20, Complex/Challenging 10 as the target proportions) and check whether your attempted items and, more importantly, your unseen accuracy track those proportions. Do not assume a large total is evenly spread; a bank can be deep in trauma and thin in the largest domain, the complex stable patient. Measure, do not assume.
Sample question style: what transfers, and what does not
What transfers well from a bank like this: factual recall, single-step diagnosis, drug and dose knowledge (verify specifics against the SmPC via the eMC), and isolated data interpretation. Where written banks are weaker: long, ED-realistic stems that bury the decision in noise; genuinely plausible distractors that separate the safe next step from the merely reasonable one; and management sequencing under time. The honest position is that a good bank builds the knowledge substrate but does not, by itself, build the timed prioritisation reflex. Judge the bank's stems on three axes — recall versus application, stem length and realism, and distractor plausibility — and expect a static bank to be strongest on the first and weakest on the last.
Jurisdiction and recency
Take a stratified sample across domains and check it against current UK guidance (NICE, CKS, SIGN, and the SmPC via the eMC), recording your review date. Emergency-medicine content is guidance-sensitive: sepsis, paediatric fluids, analgesia, toxicology and resuscitation all change. Any bank, including this one, can lag. Do not treat an explanation as current because it is written confidently; cross-check guidance-sensitive items against primary sources and note the date you did so.
The format gap
State it plainly: a standard written Q-bank — this one included — cannot fully prepare you for ED prioritisation across a 90-item timed paper, for the density of image and ECG interpretation, or for time-critical management decisions taken at roughly 80 seconds each. Those are properties of the whole paper under the clock, not of individual items. This is not a criticism specific to MRCEM Exam Prep; it is the boundary of the modality. You close that gap with timed mixed simulation on unseen items, not with more single items.
Duplication and contamination
In any large bank, concepts recur and some stems are near-duplicates. That is useful for spaced repetition but dangerous for self-assessment: once you have seen an item, answering it again measures recognition, not knowledge. As you approach the exam, your rising percentage on a bank you have largely completed overstates readiness. Protect a pool you never review inside the bank, and re-measure there — or on a separate unseen bank — to keep the signal honest.
Best-fit matrix
| Use case | Fit |
|---|---|
| Foundation building | Strong — volume plus written tutorials |
| First pass / breadth | Strong |
| Second (measurement) bank | Weak — too much overlap once used for learning |
| Retake, targeted repair | Moderate — good for content gaps, less so for pacing |
| Final timed simulation | Weak — use unseen mixed blocks instead |
Worked example: a seven-day plan for a busy trainee
Days 1–5: one defined job — domain-filtered first-pass blocks on MRCEM Exam Prep in your two weakest domains, 30–40 items a day around shifts, error-coding each miss with a single corrective action rather than transcribing the explanation. Day 6: a timed mixed block on unseen items in iatroX to measure whether the week's learning transferred — this is measurement, not more teaching, and it makes no claim about any adaptive algorithm. Day 7: review only the transfer failures and the high-confidence errors; rest. The division of labour is deliberate: the bank does breadth, the unseen block issues the readiness signal.
Reading your results: what the domain table is telling you
When you lay your MRCEM Exam Prep performance against the weighting, read three signals rather than one. Read the gap between your seen and unseen accuracy: a large gap means your bank percentage is recognition, and the unseen number is the real one. Read the shape across domains: a flat, adequate profile is safer than a spiky one with a strong average, because the paper samples every domain and a single floored domain can sink an otherwise good candidate. Read the error types within a weak domain: a domain that is weak from knowledge gaps needs the written tutorials, whereas a domain that is weak from misread stems under time needs timed practice, not more reading. The same low number can have two opposite fixes.
Where an integrated tutor helps — and where it does not
Some MRCEM banks bundle an AI tutor or explanation assistant; MRCEM Exam Prep, on the pages checked, advertises written tutorials rather than an adaptive tutor. Neither is inherently better. An explanation layer helps most when your errors are knowledge gaps and you need the reasoning made explicit; it helps least when your errors are pacing or stem-misreading, which are trained by doing timed blocks rather than by reading more. If you are comparing banks on their tutor, classify your error types first and buy the feature that fixes your actual failure — and hold any automated explanation to the same standard as any other source, checking it against current guidance before you trust it. The pillar on auditing an AI tutor sets out how to test grounding and answer-leakage.
A second reading: when your unseen score is the higher one
Occasionally the pattern reverses: a candidate's unseen block is strong while the completed-bank percentage is higher still. That is not reassurance — it means the bank has been learned rather than the subject, and the two numbers will converge downward on exam day. The corrective is to stop the primary bank, leave any remaining items untouched, and spend the final fortnight on unseen mixed simulation only, so that the number you are watching is the one the exam will actually produce.
Three mistakes this audit is designed to stop
The first is buying on catalogue size: 2,600 items is only an advantage if they are distributed across the weighting and you actually work and review them, and a smaller fully-worked bank beats a giant half-worked one. The second is trusting a rising percentage on a completed bank — near the exam that number is recognition, and it will tell you that you are ready when you are not. The third is treating explanations as current: emergency-medicine guidance moves, so cross-check guidance-sensitive items against primary UK sources and date what you find. Each mistake has the same antidote — measure on unseen, timed items and read the result by domain, not as a single headline figure.
Decision checklist
- Continue if: first pass incomplete, unseen accuracy rising, domains moving toward floor.
- Supplement if: one domain or modality (ECGs, paediatric dosing) stays weak — add a targeted tool, not another whole bank.
- Switch if: content reads stale against current guidance on repeated cross-checks, or you have completed enough that percentages are now recognition.
- Stop new questions if: all domains at floor on unseen timed blocks, pace stable, high-confidence errors low — move to simulation and rest.
Base every branch on a measured gap, not novelty or sunk cost.
FAQ
Is MRCEM Exam Prep enough for MRCEM SBA on its own? For a first-attempt candidate with solid clinical exposure it can be a sufficient primary bank for breadth and explanations, but "enough" depends on adding two things it does not itself provide: the official RCEM sample material for format calibration, and unseen, timed measurement so your readiness signal is not just recognition of items you have already worked. As a standalone that you complete and re-score, it will flatter you near the exam.
Which MRCEM SBA component does MRCEM Exam Prep not reproduce well? The whole-paper, timed prioritisation component — 90 items in two hours, roughly 80 seconds each, with realistic distractors and dense image and ECG interpretation. Like any static single-item bank, it trains knowledge and isolated reasoning better than it trains sustained pacing and mixed-domain triage. Reproduce that with timed mixed blocks, not with more single questions.
How many MRCEM Exam Prep questions should I complete per day for MRCEM SBA? There is no universal number; anchor it to review quality, not volume. A common sustainable pattern for a working trainee is 30–40 items on shift days and a larger mixed block on days off, with every miss error-coded and given one corrective action. If you are completing questions faster than you can genuinely review them, lower the count — un-reviewed volume inflates your percentage without improving transfer.
When should I stop using MRCEM Exam Prep and move to mixed mocks? Switch when your domain floors are met on unseen items even if the bank is not finished. Completion is not the criterion; unseen, timed performance is. Most candidates should move the emphasis to mixed timed papers in the final two to three weeks, using the bank only to patch a specific, measured gap.
How should I combine MRCEM Exam Prep with iatroX without duplicating practice? Assign each a non-overlapping job: MRCEM Exam Prep for first-pass breadth and written explanations; iatroX for unseen, timed measurement and for opening a missed concept in the Socratic Tutor to test the underlying principle on a fresh item. Do not re-answer the same questions in two places — that duplicates exposure and corrupts your readiness signal. The two-Q-bank rule sets out the split; make no assumptions about proprietary algorithms on either side.
Editorial notes and references
Written by Dr Kolawole Tytler, NHS GP and founder of iatroX. Last checked 19 July 2026. Question counts, prices and features attributed to MRCEM Exam Prep are vendor-reported as at that date and change; verify on mrcemexamprep.net. Disclosure: iatroX operates a competing MRCEM SBA bank and Socratic Tutor; this audit confines iatroX's role to unseen measurement and transfer practice — jobs a first-pass content bank does not claim — and is positive about MRCEM Exam Prep where its volume and price warrant. Corrections via the feedback route on iatrox.com.
References: RCEM MRCEM exams and the MRCEM SBA Regulations and Information pack with sample questions (rcem.ac.uk); MRCEM Exam Prep product pages (mrcemexamprep.net); NICE/CKS/SIGN and the SmPC via the eMC for clinical content; iatroX MRCEM SBA bank, comparison hub, the two-Q-bank rule and "Your Q-Bank Percentage Is Not Your Exam Score".