If you have failed the UKMLA Applied Knowledge Test, or a finals sitting mapped to it, the most important week is the first one — and it should be diagnostic, not frantic. Do not buy three new resources, do not restart your notes from page one, and do not redo your old question bank from question one. Spend seven days working out which failure mode cost you the marks, then rebuild around that single finding.
Failing the UKMLA does not mean you are unsafe or incapable. The pass standard is set at the level of a doctor completing their first foundation year, and most candidates who fall short have not run out of knowledge — they have run a study loop that never identified the weak presentations early enough to fix them. The AKT now draws on the MLA content map of roughly 430 conditions organised around clinical presentations, and the question style has shifted firmly towards longer vignettes that test reasoning rather than isolated recall. A revision method that worked for recall-style finals can quietly under-prepare you for that shift.
This guide deals with the knowledge component. The CPSA (the OSCE element) is a separate skill set with its own remediation, and if that is where you came unstuck, the priorities below will not be the right ones.
Do not restart blindly
The single most common post-failure mistake is to treat the result as evidence that you knew nothing, and to respond by burning the week on a panic-buy of new material. The opposite is true. You already hold most of the knowledge; the task is to find the specific holes and close them. The first 48 hours should be spent gathering information, not consuming content.
The failure modes that actually cost marks
Almost every UKMLA failure maps to one of a small number of patterns. Identifying yours is the entire point of week one.
| Failure mode | What it looks like | How to fix it |
|---|---|---|
| Content-map gap | A whole presentation you never revised properly | Blueprint-led practice on that presentation, then a short source review |
| Recognition bias | You "knew" the answer only once you saw the options | Free recall before looking at options; fresh questions, not repeats |
| Next-step error | You pick the diagnosis when the stem asks for management | Highlight the command word in every stem before answering |
| Guideline drift | You are applying out-of-date thresholds | Check current NICE/CKS guidance on the specific decision |
| Timing collapse | The last 30 questions are rushed and wrong | One-minute pacing rule and timed blocks |
| Emotional avoidance | You stop testing yourself after the result | Small, low-friction daily diagnostic blocks |
How to read what went wrong without a detailed score report
UK graduates sitting the AKT through their school, and IMGs sitting it via PLAB 1, do not always receive a granular breakdown. You can still reconstruct the picture. List every topic you remember finding hard, every question where you second-guessed, and whether your accuracy fell in the final third of the paper. Three honest data points — weak presentations, recognition versus recall, and pacing — tell you more than a percentage.
Should you change your question bank?
Usually not, at least not yet. If you completed less than 60% of any bank, your problem is coverage, not the platform — finish it. If you completed most of a bank and still failed, the issue is how you reviewed, not the questions. Switching resources at that point often just resets the recognition problem on new material.
The resources worth using honestly
There are several strong options, and the candid answer is that no single one is sufficient on its own. PassMedicine is a serious choice for breadth and high-volume single-best-answer exposure. Quesmed is well regarded for harder multi-step questions and its undergraduate and OSCE ecosystem. Pastest is a mature, exam-style bank. For the CPSA, Geeky Medics plus structured peer practice is the standard combination. Whichever you use, the GMC's official practice materials are essential, because they are written by the people who set the standard.
The mistake is rarely the platform. It is doing questions without a diagnostic review loop attached to them.
Where iatroX fits
iatroX is built to be the adaptive layer that sits on top of whatever volume bank you use. Its UKMLA bank is free and mapped to the MLA content map, and the engine uses semantic adaptive learning — it surfaces related weaknesses across topic boundaries rather than only reporting tag-level scores, so it tends to find the presentations you would not have thought to revisit. Incorrect items return at spaced intervals to interrupt the forgetting curve.
The part most relevant after a failure is the Socratic Tutor. Rather than handing you the answer, it asks a question back, works out where your reasoning went wrong, and only then resolves the misconception against the source and the official explanation. This matters because the evidence on AI study tools is now reasonably clear: in a randomised study of around a thousand students, Bastani and colleagues (PNAS, 2025) found that an answer-giving assistant left learners measurably worse on a later unaided exam, while a version that gave hints rather than answers removed that harm entirely. A question-first tutor is designed around that finding. When you are short of time, a "just explain it" override is there.
For guideline thresholds specifically — the source of so many next-step errors — Ask iatroX answers from a vetted UK corpus (NICE, CKS, SIGN, SmPC/eMC, NHS content) with citations, rather than generating from training data.
Your seven-day recovery plan
Day 1. Decompress. Collect any feedback available, note where timing hurt, and resist catastrophic conclusions. Day 2. Classify the failure using the table above. Be honest about recognition versus recall. Day 3. Run a diagnostic block on the iatroX UKMLA bank and let the dashboard show you weak presentations. Day 4. Compare your prior resource use against how you reviewed. Volume was probably not the problem. Day 5. Build a missed-presentation map: chest pain, abdominal pain, collapse, breathlessness, headache, jaundice, fever, bleeding in pregnancy, paediatrics, safeguarding. Day 6. Draft a two-week rescue plan: 60–100 adaptive questions a day, a Tutor review of every miss, and no passive reading unless a missed question triggers it. Day 7. Sit one timed mini-mock and compare your performance by presentation, not only by percentage.
The two-to-six-week resit plan
Weeks one to two are diagnostic and weak-presentation focused. Weeks three to four broaden to full content-map coverage of anything that turned up thin. Weeks five to six are timed mocks under realistic conditions, with a light final review. Throughout, keep the daily loop tight: questions first, a debrief on every miss, a guideline check where a threshold was the problem, and spaced repetition to keep recovered topics warm.
A short FAQ
Is the UKMLA harder than the old PLAB? The standard is the same foundation-year level, but the question style now rewards reasoning over recall, and the content map is broader. Candidates who prepare for reasoning find it comparable.
Should I delay my resit to revise more? Only if your diagnosis shows a genuine coverage gap. If the problem was review quality or pacing, more time without a changed method tends not to help.
Do I need a paid bank to pass? No. A free adaptive bank plus the GMC's official materials and a disciplined review loop is a credible route.