This is for students and international graduates preparing for the UK Medical Licensing Assessment who want to know what a question-bank percentage is not covering. The UKMLA is two assessments, not one: an applied knowledge test and a separate clinical and professional skills assessment sat as an OSCE. A bank can rehearse the first well and cannot touch the second — and even for the knowledge test, a percentage cannot see whether you can trace an answer to current UK guidance or whether you have actually covered the GMC content map. Those are the gaps, and only some of them belong to a Q-bank at all.
What a completion percentage shows for UKMLA — and what it hides
As of 19 July 2026, several banks cover the UKMLA applied knowledge test with single-best-answer items mapped to the GMC content areas; treat any question count or price as vendor-reported and confirm it on the product page. What none of them covers is the clinical and professional skills assessment — the OSCE half — because that is an observed, examiner-marked performance, not a multiple-choice test. This is the honest headline: a bank percentage, however high, measures at most half of the UKMLA, and the half it cannot measure is the one where consultation skills, examination technique and safe practice are judged directly.
Within the knowledge half, the percentage still hides two things. It cannot see whether you know why an answer is correct in UK terms — its basis in NICE, CKS, SIGN, the SmPC or NHS guidance — which matters both for the exam and for the graduate you are about to become. And it cannot see whether your attempted items actually span the GMC content map, including the professional, ethical and legal areas that banks under-represent. The rest of this article separates the three gaps and gives each an honest owner.
The official UKMLA format, mapped
The UKMLA has two components, assessed separately:
| Component | Format | Who sits which | Anchor |
|---|---|---|---|
| Applied Knowledge Test (AKT) | Single-best-answer MCQ, ~1 minute per item | MSC AKT for UK students; PLAB 1 (180 SBA, 3 hours) for IMGs | GMC MLA content map |
| Clinical & Professional Skills Assessment (CPSA) | OSCE-style stations, examiner-marked | Run by UK medical schools (and the equivalent for IMGs) | Consultation, examination, procedures, safe practice |
Both components are built on the GMC's MLA content map, which sets out the areas of clinical practice, the presentations and the conditions a Foundation-ready doctor is expected to manage. The content map was updated in January 2026 and applies from September 2026, so candidates sitting from autumn 2026 should map their revision to the current version — confirm which version applies to your sitting on the GMC site. The AKT tests whether you know; the CPSA tests whether you can do — and a Q-bank speaks only to the first.
Knowledge versus performance: what a correct answer proves
A correct AKT answer proves you can select the right option under about a minute of time pressure. It does not prove you can take a history that reaches that diagnosis, examine safely, explain the plan to a worried patient, or perform the procedure — all of which the CPSA marks directly and none of which a click can demonstrate. Nor does a correct click prove you know the UK basis for the answer: a candidate trained elsewhere may pick the internationally common option and lose nothing on the AKT while carrying a home-country default that a UK examiner, and UK practice, would not accept. This is the separation to hold onto, and it is why your Q-bank percentage is not your exam score understates the problem here: for the UKMLA the percentage is not just an imperfect readiness signal, it is a signal for only one of two assessments.
The three things a percentage cannot see
For each, here is the behaviour, the practice task, the honest feedback source, and the exit standard. One of the three is explicitly not a Q-bank job.
| Skill | Observable behaviour | Deliberate-practice task | Feedback source | Exit standard |
|---|---|---|---|---|
| CPSA / OSCE performance | Structured consultation, safe examination, clear explanation | Timed role-played stations with peers or simulated patients, marked to the domains | Examiner / clinician / tutor — not AI | Consistent station passes against your school's rubric |
| UK guideline provenance | You can state the UK basis and its recency for a management choice | For every miss, verify the source (NICE, CKS, SIGN, SmPC, NHS) and its date | Primary UK guidance; a citation-first check | You can cite the current UK basis for high-yield presentations |
| MLA content-map coverage | Attempted items span all areas, including ethics, law and prescribing safety | Map items to the content map; drive up empty cells | The GMC MLA content map itself | No blank cells in high-yield or professional areas |
CPSA / OSCE performance. This is the clearest limit of any question bank, and it should be stated plainly: no Q-bank, and no AI, prepares you for the CPSA. It is an observed performance — history-taking, examination, communication, procedures, data-gathering and safe practice — judged by examiners against a rubric, and it is trained the only way performances are trained: by doing timed stations with peers or simulated patients and getting human feedback. iatroX does not simulate the CPSA and does not claim to; its only honest contribution to this half is the underlying clinical knowledge, because you cannot manage in a station what you cannot recognise. For everything else in the CPSA, book the consultation practice.
UK guideline provenance. For every AKT miss, do not just read the explanation — verify the UK source and its recency: which NICE, CKS or SIGN guidance, which SmPC point, which NHS pathway, and when it was last updated. This is the one gap where a citation-first tool is a genuine fit: Ask iatroX is built to return an answer with its UK source attached, which is exactly the habit an internationally trained candidate needs to convert a home-country default into a UK-calibrated rule.
MLA content-map coverage. Completion of a bank is not coverage of the content map. Map your attempted items to the areas of clinical practice, presentations and conditions, and pay particular attention to the professional, ethical and legal cells — capacity, consent, safeguarding, prescribing safety — which banks reliably under-weight. The content map, in its current version, is the checklist; the completion-is-not-coverage guide shows how to turn it into a matrix with no empty high-yield cells.
A four-week modality ladder
Two tracks run in parallel, because the two components are trained differently.
| Week | Rung | Knowledge track (AKT) | Skills track (CPSA) |
|---|---|---|---|
| 1 | Isolated skill | Single-topic drills + guideline-provenance checks | One examination or communication skill at a time |
| 2 | Coached case | Mixed items, reasoning before reveal; misses sourced to UK guidance | Role-played stations with feedback from a tutor or peer |
| 3 | Timed integrated | Full timed AKT blocks at ~1 minute per item | Timed OSCE circuits against the domains |
| 4 | Unseen simulation | A fresh, content-map-proportioned, timed AKT block | A mock OSCE with fresh stations and a real examiner |
The knowledge ladder can use a preserved unseen reserve for week four; the skills ladder needs human examiners throughout, and no amount of AKT practice substitutes for a single marked mock station.
When AI feedback helps, when it misleads, and when you need a clinician
For the AKT, AI feedback helps on defined jobs: explaining a thin distractor, generating variant stems to test transfer, and — its strongest use here — attaching a current UK source to a management point through a citation-first tool, so guideline provenance becomes a habit rather than an afterthought. It misleads whenever it is asked to judge a performance: an AI cannot watch your consultation, cannot confirm your examination was safe, and cannot mark a station, and any score it offers for those is fabricated confidence. Read how to calibrate AI feedback before trusting an automated grade. For the entire CPSA, and for verifying that your consultation and examination meet the standard, the required feedback source is a clinician or examiner — this is the part of the UKMLA where a human is not optional.
A balanced task matrix
For the AKT, cross the content-map areas with the task the item demands so you are not only revising familiar recall; track CPSA domains separately, because they are marked by a person, not a percentage.
| Content-map area | Recall | Diagnosis from data | Next-step management | CPSA domain (track separately) |
|---|---|---|---|---|
| Acute & emergency | ✓ | ✓ | ✓ | Data-gathering & examination |
| Cardiorespiratory | ✓ | ✓ | ✓ | Clinical management |
| Mental health | ✓ | ✓ | ✓ | Communication & consent |
| Child health | ✓ | ✓ | ✓ | Safe practice / safeguarding |
| Ethics, law & prescribing safety | ✓ | ✓ | ✓ | Professionalism |
The bottom row is the one banks thin out and the CPSA leans on; populate it deliberately in both tracks.
A worked example: two tracks from one flattering number
Take a final-year student at 82% on their AKT bank who feels well placed. Two checks reframe it. Mapping their attempted items to the current content map shows strong cardiorespiratory and acute coverage but near-empty cells in capacity and consent, safeguarding thresholds and prescribing safety — the professional areas the bank under-samples and the January 2026 content-map update reinforced. And a single honest question — have they done any marked CPSA stations? — reveals none, because the AKT percentage felt like progress and the OSCE felt distant.
The framework produces two plans, not one. On the knowledge track, the empty professional cells get a floor and a daily allocation, and every miss is sourced to current UK guidance so provenance is trained alongside recall. On the skills track, timed role-played stations start immediately with tutor or peer feedback against the CPSA domains, building towards a mock OSCE with a real examiner. The 82% is retired as the headline; the signals that matter are content-map coverage with no empty high-yield cells, provenance the student can cite, and station passes against the rubric. No number here predicts either result — the exercise stops the AKT percentage from crowding out the half of the UKMLA it cannot measure.
Three mistakes this framework is designed to stop
The first is letting the AKT stand in for the whole UKMLA. The two assessments are separate, and a high knowledge percentage says nothing about the OSCE — so ring-fence CPSA practice from day one rather than deferring it. The second is treating completion as coverage: banks under-represent ethics, law and prescribing safety, so a "finished" bank can leave the professional content map empty; measure coverage against the map, not the bank, and keep a second unseen source for honest measurement under the two-Q-bank rule. The third, especially for international graduates, is answering correctly without knowing the UK basis, so a home-country default survives untested until an examiner or a patient finds it; make guideline provenance an explicit review category.
Continue, supplement, switch or stop
Continue with your AKT bank while unseen mixed accuracy is still climbing and your content-map coverage is still filling. Supplement with unseen timed blocks the moment your completion percentage outpaces your performance on novel questions, and supplement with provenance work whenever you can pick the answer but not its UK source. Switch or add a bank only for a measurable content-map gap — thin professional, ethical or prescribing-safety coverage — not for a slicker interface. Stop grinding the AKT bank the moment it starts crowding out CPSA practice, because for most candidates the OSCE, not the knowledge test, is the under-rehearsed half — and no percentage will warn you of that.
The bottom line
The UKMLA is two assessments, and a question bank honestly serves one of them. Use a strong bank for the applied knowledge test, measure it on unseen timed blocks mapped to the current GMC content map, and make guideline provenance a habit with a citation-first check. Then treat the CPSA as the separate, human-marked performance it is, and give it real consultation practice with examiner feedback. A bank percentage is a signal for half the exam; the other half is trained in a room with a patient and an examiner, and no software replaces that.
Frequently asked questions
How do I know whether I have covered the full UKMLA blueprint? The GMC MLA content map is the blueprint for both components; map your attempted AKT items to its areas of clinical practice, presentations and conditions, and set a floor per cell so the professional, ethical and legal areas are not empty. Confirm whether the current (January 2026) content-map version applies to your sitting, and remember that content-map coverage measures AKT readiness only — the CPSA is assessed separately.
Can one question bank be enough for UKMLA? For the AKT, a strong bank can anchor your learning, though you still need a second unseen source to measure readiness rather than recognition. But no bank covers the CPSA, which is a separate OSCE requiring observed consultation practice — so a single question bank is structurally insufficient for the UKMLA as a whole, and treating it as sufficient is the commonest planning error.
What should I measure instead of my overall Q-bank percentage for UKMLA? For the AKT, measure first-pass accuracy on unseen, timed blocks mapped to the content map, plus whether you can cite the current UK basis for high-yield presentations. For the CPSA, measure station pass rate against your school's rubric under a real examiner. These are two separate signals, and the single blended bank percentage tracks neither well.
When should I stop doing new UKMLA questions? Stop adding AKT questions when unseen timed blocks are stable across the content map and new items stop revealing gaps — and stop earlier than that if AKT drilling is displacing CPSA practice, because most candidates over-invest in the knowledge test and under-practise the OSCE. Finishing the bank is not readiness; balanced coverage plus marked station practice is.
Which UKMLA resource should I use for my weakest component? Match the resource to the component. Weak AKT knowledge: a broad bank governed by the content map. Weak guideline provenance: primary UK guidance and a citation-first check such as Ask iatroX. Weak CPSA performance: timed consultation practice with examiner or tutor feedback — explicitly not a question bank, because you cannot rehearse an observed performance by clicking options.
Editorial notes and references
Written by Dr Kolawole Tytler, NHS GP and founder of iatroX. Last checked 19 July 2026; any bank question counts or prices mentioned are vendor-reported — verify them on the relevant product page. UKMLA components and the MLA content map (updated January 2026, applying from September 2026) are per the GMC. Disclosure: iatroX operates a competing UKMLA applied-knowledge bank, and its role here is confined to the AKT half — unseen, timed measurement and citation-first guideline checks — not the CPSA, which is an examiner-marked OSCE iatroX does not simulate. Corrections via the feedback route on iatrox.com. References: GMC Medical Licensing Assessment and content map (gmc-uk.org); the iatroX comparison hub; related reading: why your Q-bank percentage is not your exam score and completion is not coverage.