This is a decision tree, not a league table. The right UKMLA resource depends on your starting profile, the weeks left before your diet and your budget — not on which bank a forum thread calls "the best". Below, four measurable inputs route you to a minimum stack: your blueprint-coverage gap, your unseen-score trend, your weeks-to-exam and your dominant error type. iatroX is used here for one narrow job — the unseen baseline that tells you which branch you are actually standing on.
Start with the exam, not the shop
The UK Medical Licensing Assessment (UKMLA) has two parts. The applied knowledge test is delivered as the MSC AKT for UK medical students and as PLAB 1 for international medical graduates; both are single-best-answer papers built on the GMC's MLA content map. The second part is the Clinical and Professional Skills Assessment (CPSA), an OSCE-style clinical examination that no multiple-choice bank can reproduce. The AKT runs at roughly one minute per item, so pace is a genuine constraint, not an afterthought.
The content map is the thing to buy resources against. The GMC updated it for the 2026 cycle, and the updated map applies to every MLA assessment sat from September 2026 onwards; candidates sitting before then are still assessed against the previous (2019) map, so check which version governs your diet. The map is organised around areas of clinical practice, areas of professional knowledge, patient presentations and conditions — the A–Z lists run to over 200 presentations and over 400 conditions in the 2026 version (confirm the current counts on the GMC page). Every resource decision below is downstream of "does this help me cover that map and perform under those conditions".
The four inputs that actually route your decision
Ignore novelty and sunk cost. Four measurable inputs decide your branch.
- Blueprint-coverage gap. The proportion of content-map areas of clinical practice where you have attempted few questions or score below your personal target. A gap over roughly 30% of areas is a coverage problem, not a polish problem.
- Unseen-score trend. Your accuracy on fresh, mixed, timed blocks you have not seen before, tracked over two to three weeks. A flat or falling trend under exam conditions matters far more than a rising "revised bank" percentage.
- Weeks-to-exam. Under four weeks, four to twelve weeks, or more than twelve weeks. This sets how much you can add versus what you must protect.
- Dominant error type. Sort a sample of your wrong answers into knowledge gap, application or reasoning error, pacing (ran out of time), or careless slip. The most common category tells you whether you need content, transfer practice or timing drills — and those need different tools.
You cannot read three of these four from a single bank's home screen. That is why the first action in this tree is a fresh, unseen baseline, not a purchase.
The minimum viable stack
Most candidates need less than they are sold. A defensible UKMLA stack is:
- One primary Q-bank worked to genuine coverage across the content map — not two banks skimmed.
- Official calibration material — the GMC/Medical Schools Council sample and specimen content — used to check that your practice questions match the real format and standard.
- One teaching or reference source, only where a domain is genuinely weak — a notes product, video series or the clinical AI you already use to resolve misses.
- One modality tool where relevant — for the CPSA, structured station practice with a real observer; a Q-bank does not count here.
Add a second bank only under the two-Q-bank rule: a defined, non-overlapping job (for example, unseen measurement) rather than "more questions". Adding volume you never analyse is the most common way candidates waste both money and calibration.
Budget bands
Verify every price on the day you buy — vendors change plans and currencies without notice.
- Free / low-cost. iatroX's free UK-core banks plus the GMC's own materials cover the baseline-and-calibration jobs at no cost. Geeky Medics offers a large volume of free UKMLA teaching content around a paid MLA AKT bank.
- One premium resource. A single mainstream subscription (for example Quesmed, Pastest, PassMedicine or UWorld) worked properly is enough for most first-attempt candidates. Confirm the current price and access window on the product page.
- Comprehensive stack. A premium primary bank, an official calibration set and a targeted teaching source. Beyond this, extra spend rarely buys extra readiness — it buys unanalysed questions.
Time bands, and what to omit
The discipline is deciding what to cut, not what to add.
- Under four weeks. Protect timed mixed blocks and error analysis. Omit new content courses and any second bank. One bank, worked in timed sets, plus the official sample for calibration.
- Four to twelve weeks. The standard window. One primary bank to coverage, weekly mixed timed blocks, targeted teaching only for your two weakest areas. Add CPSA station practice on a separate track.
- More than twelve weeks. You can afford a content-first phase and a broader teaching source, but keep an unseen block running weekly from the start so your trend line has history. Do not let a long runway become a permanent "revising" loop with no measured transfer.
Decision matrix: platform to the one job it does best
Summarised deliberately — for the detailed evidence, follow the linked child audits rather than re-reading a description here.
| Resource | The one job it does best | Where it is weak | Detail |
|---|---|---|---|
| GMC / MSC official material | Format and standard calibration | Finite volume | GMC content map (References) |
| Geeky Medics | Teaching and foundation, mock papers | No adaptive or AI analytics | Vendor page |
| Quesmed | Volume plus CPSA/OSCE station practice | Percentage inflation on repeats | Comparison hub |
| Pastest | Volume and exam-style realism | Cost of full access | Comparison hub |
| PassMedicine | Concise notes plus value | Lean analytics | Comparison hub |
| UWorld | Explanation depth for reasoning errors | Price; US-origin framing | Comparison hub |
| MedRevisions | IMG/PLAB volume plus analytics | AKT-only; no CPSA | MedRevisions audit |
| MedSnapp | Gamified active recall from your own notes | Not blueprint-calibrated; no PLAB 1 | MedSnapp audit |
| MediWord | Recall foundation plus notes | Recall-heavy; limited applied transfer | MediWord audit |
| iatroX | Unseen, mixed, timed measurement | Not a CPSA simulator | UKMLA bank |
The decision table: condition to branch
| If your situation is… | Then your next branch is… | Primary measurable trigger |
|---|---|---|
| First attempt, over eight weeks out, wide coverage gap | One premium primary bank, content-first, weekly unseen block | Coverage gap > 30% of areas |
| Strong revised percentage but failing unseen blocks | Stop re-reading; switch to unseen transfer measurement | Unseen trend flat/falling |
| IMG balancing home practice with UK conventions | Primary bank for volume, plus deliberate UK-guideline localisation | Errors clustered in UK-context items |
| Weak foundations across several systems | Teaching source first, bank second, no second bank | Knowledge-gap errors dominate |
| Strong knowledge, runs out of time | Timed mixed blocks only; drop new content | Pacing errors dominate |
| Under four weeks, moderate gap | One bank in timed sets plus official calibration | Weeks-to-exam < 4 |
| Retake after a near miss | Diagnose error type first, then target that one input | Dominant error type unknown |
Three worked candidate profiles
Aisha — UK final-year student, first attempt, ten weeks out. Baseline on a fresh unseen block shows 68% overall but a coverage gap in paediatrics, psychiatry and clinical ethics. Branch: one premium primary bank worked system by system, weekly 90-item timed mixed block, targeted teaching for the three weak areas only. Weekly allocation: 4 days content-and-questions, 1 day mixed timed block, 1 day error analysis, 1 day rest. Exit criterion: unseen mixed accuracy stable above her faculty's historical pass region for three consecutive weekly blocks, with no content-map area left below 60%.
Priya — international graduate, PLAB 1 route, twelve weeks out. Strong disease knowledge, but errors cluster in UK-specific management sequencing, referral pathways and professional-framing items. Branch: a high-volume bank (for example MedRevisions or Quesmed) for coverage, plus a deliberate localisation loop where every missed UK-context item is re-derived against current UK guidance. Weekly allocation: 5 short question sets, 1 UK-guideline localisation session on the week's misses, 1 unseen timed block. Exit criterion: UK-context error rate halved and unseen trend rising across four weeks.
Tom — retake candidate, six weeks out. Revised-bank percentage was 82% first time, yet he failed. Error sort shows pacing and application errors, not knowledge gaps. Branch: stop new content and repeated familiar questions; move entirely to timed mixed unseen blocks with strict per-item pacing and post-block error analysis. Weekly allocation: 3 timed blocks, 3 analysis-and-retest sessions, 1 rest day. Exit criterion: completes full-length timed papers within time with careless-error rate under 5% and a stable unseen trend.
How to weigh evidence about resources
Use a hierarchy. Official GMC and MSC material is first for format and standard. Primary UK guidance (NICE, CKS, SIGN, the SmPC/eMC and NHS content) is first for what is clinically correct. Vendor pages are acceptable only for product facts — count, price, features — and should be labelled vendor-reported and dated. Independent testing and your own unseen scores are what tell you whether a resource actually improves transfer. Never let a marketing claim outrank your measured trend.
Bottom line
There is no universal best UKMLA resource; there is a best next action for your four inputs. Take an unseen baseline, sort your errors, count your weeks and estimate your coverage gap. Those numbers pick your branch, your budget band and your stack. The banks are largely interchangeable for volume — what is not interchangeable is disciplined coverage against the content map and honest, unseen measurement of transfer.
Frequently asked questions
How do I know whether I have covered the full UKMLA blueprint? Map your attempts, not your feelings. Build a simple matrix of the content map's areas of clinical practice, key presentations and conditions, and record attempted volume and unseen accuracy in each cell; an area with few attempts or low unseen accuracy is uncovered regardless of your overall percentage. Our blueprint-coverage-matrix method walks through this, and the GMC content map is the authoritative list to build it from.
Can one question bank be enough for UKMLA? For most first-attempt candidates, yes — one primary bank worked to genuine coverage, paired with official calibration material and separate CPSA practice, is sufficient. A second bank earns its place only under a defined, non-overlapping job such as unseen measurement; adding a second bank for "more questions" duplicates content and corrupts your percentage without improving readiness.
What should I measure instead of my overall Q-bank percentage for UKMLA? Measure your accuracy on fresh, unseen, timed, mixed blocks and watch its trend over time, broken down by content-map area. A revised-bank percentage is inflated by recognition and reveals little; the unseen trend is the closest proxy you have to exam-day performance. Our guide to why a Q-bank percentage is not your exam score explains the distinction in full.
When should I stop doing new UKMLA questions? When new questions stop changing your decisions. If a fresh block surfaces no new content-map gaps and your unseen trend has plateaued at a safe level, further new volume is low-value; your remaining time is better spent on timed full-length papers, error analysis and CPSA practice. Doing more questions to feel productive is not the same as closing a measured gap.
Which UKMLA resource should I use for my weakest component? It depends which component is weakest. For a knowledge gap, a teaching or reference source plus targeted questions in that area; for an applied-reasoning gap, a bank with strong explanations (used to re-derive, not re-read); for pacing, timed mixed blocks; and for the CPSA, structured station practice with a real observer, because no MCQ bank — iatroX included — reproduces the OSCE. Diagnose the component with an error sort first, then choose the tool that matches it.
Editorial notes and references
Written by Dr Kolawole Tytler, NHS GP and founder of iatroX. Last checked 19 July 2026. Vendor-reported figures (question counts, prices, features) were taken from the named product pages on that date and change without notice — confirm the current figures before you buy. Disclosure: iatroX operates a UKMLA question bank and therefore competes with several resources named here; this article confines iatroX to jobs the other products do not claim — a free unseen baseline and mixed transfer measurement — and links out to independent detail for the rest. Corrections are welcome via the feedback route on iatrox.com.
References: GMC, Medical Licensing Assessment and MLA content map (gmc-uk.org); vendor product pages (Geeky Medics, Quesmed, Pastest, PassMedicine, UWorld, MedRevisions, MedSnapp, MediWord); iatroX, "Your Q-Bank Percentage Is Not Your Exam Score" (https://www.iatrox.com/blog/qbank-percentage-not-your-exam-score); iatroX, "Question-bank completion is not coverage" (https://www.iatrox.com/blog/question-bank-completion-is-not-coverage-how-to-build-a-blueprint-coverage-matrix-for-any-medical-exam); iatroX, "The Two-Q-Bank Rule" (https://www.iatrox.com/blog/the-two-q-bank-rule-how-to-add-a-second-bank-without-duplicating-questions-or-destroying-calibration); iatroX comparison hub (https://www.iatrox.com/compare).