This is for applicants preparing for the Multi-Specialty Recruitment Assessment who want to know what a bank percentage is not measuring. The MSRA is two structurally different papers whose combined score feeds a national ranking, and four things decide that ranking which an ordinary bank cannot assess: whether your situational-judgement responses match the official consensus, where your effort actually moves your rank, whether your pace survives two very different papers, and whether you can switch cold between clinical domains item to item. Each is trainable, but only some of it belongs to a Q-bank — and the paper that most defines your professionalism score is not one of them.
What a completion percentage shows for MSRA — and what it hides
As of 19 July 2026, several commercial banks cover the MSRA with situational-judgement scenarios and clinical single-best-answer items; treat their scenario counts, scoring keys and prices as vendor-reported and confirm them on the product page. Two cautions apply before you trust any of it. First, the assessment is norm-referenced: your combined score is used to rank you within an applicant pool for shortlisting and, in several specialties, as part of the final appointment score, so this is not a simple pass/fail hurdle where a comfortable percentage means done. Second, and more important, a third-party bank cannot reproduce the scoring that decides half your result.
A completion percentage on an MSRA bank tells you how many of that bank's items you have attempted. It cannot tell you whether your judgement on the Professional Dilemmas paper matches the panel consensus that actually scores it, whether the marks you are chasing are the ones that move your rank, or whether you can hold pace across a 95-minute judgement paper and a sub-minute clinical paper on the same afternoon. Those are the constraints the ranking is built on, and they are exactly what the percentage cannot see.
The official MSRA format, mapped
The MSRA runs for 170 minutes across two papers, sat in one sitting:
| Paper | Time | Items | Nature | Scoring |
|---|---|---|---|---|
| Professional Dilemmas (PD) | 95 min | ~50 situational-judgement scenarios | Ranking and multiple-response judgement items | Consensus-scored against an expert panel key |
| Clinical Problem Solving (CPS) | 75 min | ~86 SBA / EMQ | Clinical knowledge, primary-care-weighted | Best-answer, sub-minute pace |
The two papers reward different things. CPS is a knowledge-and-reasoning paper spanning general medicine, surgery, paediatrics, psychiatry, obstetrics and gynaecology and more, weighted towards primary-care presentations, run at roughly 52 seconds an item. PD is a professionalism paper: scenarios probing attributes such as coping with pressure, communication, patient focus, integrity and teamworking, mapped to the standards in the GMC's Good Medical Practice, and scored not against a single right answer but against a consensus key derived from an expert panel. Confirm the current item numbers and any format detail in the official applicant guidance, because recruitment specifications are revised between cycles.
Knowledge versus performance: what a correct answer proves
On CPS, a correct answer in untimed single-topic revision proves you can recognise the right management for that presentation with priming and time. It does not prove you can retrieve it at 52 seconds an item, cold, when the previous question was psychiatry and the next is obstetrics. On PD, the separation is sharper still: a "correct" answer on a commercial SJT proves only that you matched that vendor's key. If the vendor's rationale diverges from the official consensus, a high bank score can actively mis-calibrate you — you are rehearsing a judgement the real panel would mark down. This is why the your-percentage-is-not-your-score caveat matters twice over for the MSRA: on CPS the percentage overstates timed performance, and on PD it can be measuring the wrong target entirely.
The four things a percentage cannot see
For each, here is the behaviour to watch, the practice task, where trustworthy feedback comes from, and the standard to move on. Note that for one of them, the honest feedback source is not any commercial product.
| Skill | Observable behaviour | Deliberate-practice task | Feedback source | Exit standard |
|---|---|---|---|---|
| SJT calibration | Your rankings match the consensus rationale, not a house key | Work official practice scenarios; justify each ranking against a professional principle | Official consensus-scored material + Good Medical Practice | Your orderings track the official rationale, and you can explain why |
| Rank strategy | Effort goes to the highest marginal-yield paper | Estimate where a marginal hour moves your position; protect the coachable paper | Unseen timed performance relative to a cohort, where available | Extra questions no longer move your unseen score |
| Time pressure | Accuracy does not collapse in the last items of either paper | Full-length timed CPS and PD, back to back | First-pass accuracy by quartile; pace log | No decay across a simulated two-paper sitting |
| Mixed-domain discrimination | Cold-switch accuracy matches single-topic accuracy | Interleaved, unseen, primary-care-weighted CPS blocks | Per-domain accuracy on mixed blocks | Mixed accuracy within a few points of single-topic |
SJT calibration. This is the honest limit of any Q-bank. The Professional Dilemmas paper is consensus-scored, and only official practice material carries the real scoring logic. A commercial SJT bank can familiarise you with the format, but drilling it hard to a third-party key risks entrenching orderings the panel would not reward. The reliable route is the official practice paper, worked slowly, with each ranking justified against a Good Medical Practice principle rather than a gut feeling — and iatroX does not attempt to reproduce that consensus scoring, so treat PD as a job for official material, not for us.
Rank strategy. Because the MSRA ranks rather than simply passes you, the strategic question is where a marginal hour moves your position. CPS is generally the more coachable paper — knowledge responds to work — while PD has a lower ceiling once your calibration is sound. A bank percentage says nothing about your likely rank; unseen timed performance, ideally benchmarked against a cohort, is the only signal that does.
Time pressure. The two papers demand different tempos, and the switch between analytical clinical reasoning and reflective professional judgement is itself tiring. Rehearse them back to back, timed, and log accuracy by quartile so end-of-paper decay is visible.
Mixed-domain discrimination. CPS never gives you forty psychiatry items in a row. Practise interleaved, unseen, primary-care-weighted blocks so you train the cold switch the real paper demands, and compare mixed-block accuracy with your single-topic figure to see how much priming was flattering you.
A four-week modality ladder
| Week | Rung | Clinical (CPS) | Professional (PD) |
|---|---|---|---|
| 1 | Isolated skill | Single-domain drills, untimed then timed | Read Good Medical Practice; work principles behind official scenarios |
| 2 | Coached case | Mixed items, reasoning before reveal, misses classified | Official practice scenarios with published rationale, justified aloud |
| 3 | Timed integrated | Full 86-item timed CPS blocks | Full 95-minute timed PD on official material |
| 4 | Unseen simulation | Fresh interleaved CPS block, never seen | Official practice paper sat cold, back to back with CPS |
The ladder keeps PD anchored to official material throughout, because that is the only source whose scoring reflects the real key. On CPS, hold a reserve of unseen questions for week four so the simulation genuinely tests transfer.
When AI feedback helps, when it misleads, and when you need a clinician
For CPS, AI feedback earns its place on narrow jobs: explaining a thin distractor, generating variant clinical stems to test transfer, and checking that a management point matches current UK primary-care guidance through a citation-first tool such as Ask iatroX. For PD it is actively risky. An AI has no access to the consensus key, so its confident ordering of a professionalism scenario is a guess dressed as a mark — and rehearsing to it can move you away from the panel, not towards it. Before trusting any automated judgement, read how to calibrate AI feedback. The human you need is a clinician or educational supervisor who can talk through the professional principles behind a scenario, because calibrated judgement — not a generated score — is what the PD paper actually measures.
A balanced task matrix
Plan CPS blocks against domain crossed with task, so you are not only revising familiar territory, and track PD separately by professional attribute.
| CPS domain | Recall | Diagnosis from data | Next-step management | PD attribute (track separately) |
|---|---|---|---|---|
| Acute & general medicine | ✓ | ✓ | ✓ | Coping with pressure |
| Paediatrics | ✓ | ✓ | ✓ | Communication |
| Psychiatry | ✓ | ✓ | ✓ | Patient focus |
| Obstetrics & gynaecology | ✓ | ✓ | ✓ | Integrity / probity |
| Prescribing & therapeutics | ✓ | ✓ | ✓ | Teamworking |
The primary-care weighting means the management column carries most of the marks; populate it deliberately rather than defaulting to recall items.
A worked example: reallocating from a flattering CPS score
Take an applicant four weeks out with 85% on their CPS bank and little structured PD work, who assumes the clinical paper is handled. Two tests change the picture. A fresh, interleaved, unseen CPS block comes back at 68% — the seventeen-point drop is domain-switching and pace, not knowledge, because the bank had been drilled in comfortable single-topic sets. And a slow pass through the official PD practice paper shows their instinctive orderings diverging from the consensus rationale on integrity and escalation scenarios, precisely the attributes that separate ranks.
The reallocation is straightforward once the two signals are visible. CPS work shifts from single-topic grinding to timed interleaved blocks so the cold-switch penalty shrinks, with pace logged by quartile. PD work moves onto official material worked slowly, each ranking justified against a Good Medical Practice principle until the applicant can predict the consensus rather than guess it. The 85% is retired as a target; the signals that matter are unseen CPS accuracy under time and PD orderings that track the official key. Nothing here forecasts a rank — the exercise removes the two weaknesses the bank percentage hid.
Three mistakes this framework is designed to stop
The first is treating a CPS bank percentage as readiness. Once seen, its items measure recognition, so keep a second unseen source for measurement under the two-Q-bank rule and let your signal come from cold questions. The second is over-drilling third-party SJTs: past a point, rehearsing a vendor's key entrenches judgement the official panel would mark down, so once your calibration on official material is sound, stop. The third is single-domain comfort on CPS, which builds priming the mixed paper removes and inflates your session accuracy. Each mistake yields a reassuring number and a real gap in the thing being ranked.
Continue, supplement, switch or stop
Continue with your CPS bank while unseen mixed accuracy is still climbing and your domain coverage is still filling. Supplement with unseen timed blocks the moment your completion percentage outpaces your performance on novel questions, and supplement PD with official practice material the moment your orderings and the consensus diverge. Switch banks only for a measurable clinical-coverage gap, not for a slicker interface. Stop drilling PD once you can reliably predict the consensus, and stop grinding CPS in the final fortnight once coverage and pace hold — the last weeks are for unseen simulation and calibration review, not another lap of familiar items.
The bottom line
The MSRA is decided by a national ranking, and a bank percentage is silent on everything that sets it: whether your professional judgement matches the consensus that scores PD, where your effort actually moves your position, whether your pace survives two very different papers, and whether you can switch cold between clinical domains. Use a strong CPS bank for learning, measure it on unseen timed blocks, and hand the Professional Dilemmas paper to official consensus-scored material — because that is the one part of this exam no third-party key, and no AI, can honestly rehearse.
Frequently asked questions
How do I know whether I have covered the full MSRA blueprint? There are two blueprints, not one. For CPS, map attempted-and-reviewed items to the clinical domains, weighted towards primary care, and set a floor per cell so no area is empty — the completion-is-not-coverage guide shows how. For PD, coverage means the professional attributes in Good Medical Practice, tested on official scenarios, not a count of third-party SJTs.
Can one question bank be enough for MSRA? For CPS, one strong bank can anchor your learning, but you still need a second unseen source to measure readiness rather than recognition. For PD, no commercial bank substitutes for official consensus-scored practice material, because only the official key reflects the scoring the real panel applies — so a single bank is structurally insufficient for the whole assessment.
What should I measure instead of my overall Q-bank percentage for MSRA? For CPS, measure first-pass accuracy on unseen, interleaved, timed blocks and your pace at roughly 52 seconds an item. For PD, measure how closely your orderings track the official consensus rationale. A blended bank percentage mixes a coachable knowledge paper with a judgement paper it cannot score, so it is the wrong single number to watch.
When should I stop doing new MSRA questions? Stop CPS questions when unseen timed blocks are stable across the domains and new items stop revealing gaps. Stop PD drilling earlier and more deliberately: once you can predict the official consensus, further third-party SJTs can erode rather than improve your calibration, so protect the calibration you have rather than diluting it.
Which MSRA resource should I use for my weakest component? Match the resource to the miss. Weak clinical knowledge: a broad CPS bank governed by a coverage matrix. Weak pace or domain-switching: timed, interleaved, unseen blocks. Weak professional judgement: official PD practice material and Good Medical Practice, not more third-party SJTs. Choosing by component, not brand, is the whole point.
Editorial notes and references
Written by Dr Kolawole Tytler, NHS GP and founder of iatroX. Last checked 19 July 2026; any bank scenario counts, scoring keys or prices mentioned are vendor-reported — verify them on the relevant product page. MSRA format is per the official specialty-recruitment applicant guidance. Disclosure: iatroX operates a competing MSRA clinical bank, and its role here is confined to the CPS half — unseen, timed measurement and citation-first guideline checks — not the consensus-scored Professional Dilemmas paper, which belongs to official material. Corrections via the feedback route on iatrox.com. References: MSRA applicant guidance and official practice materials (specialty recruitment / Oriel); GMC Good Medical Practice (gmc-uk.org); the iatroX comparison hub; related reading: why your Q-bank percentage is not your exam score.