Most candidates arrive at the AMC MCQ with a question bank and a percentage. Both are useful; neither tells you whether you can reliably pick the Australian answer under time pressure, unlearn a reflex from another health system, or hold pace across 150 items in one sitting. This article is for international medical graduates on the standard pathway who can already answer clinical questions but are not yet certain they can perform on the day. It names three skills ordinary MCQ practice cannot measure — Australian guideline localisation, international-graduate transition and pacing on a long computer-administered test — and gives you an observable behaviour, a practice task, a feedback source and an exit standard for each.
The exam this article assumes
The AMC MCQ examination (last checked 19 July 2026) is a computer-administered, single-best-answer test of 150 multiple-choice questions, each with one correct response from five options, delivered in one 3.5-hour session at a Pearson VUE test centre. It covers the disease process, clinical examination and diagnosis, and investigation, therapy and management, and it is the written gateway to the AMC Clinical Examination. The AMC publishes a content blueprint weighted by discipline:
| Discipline | Approximate weighting |
|---|---|
| Adult Health – Medicine | 30% |
| Adult Health – Surgery | 20% |
| Women's Health – Obstetrics and Gynaecology | 25% |
| Child Health | 12.5% |
| Mental Health | 12.5% |
| Population Health and Ethics | Distributed across the settings above |
One honesty note before you plan around it. The AMC names this a "Computer Adaptive Test (CAT)", but the public examination page does not, on that page, set out a difficulty-adapting algorithm; it describes a fixed-length, 150-item, computer-administered examination. For planning purposes treat it as a fixed-length computer-administered MCQ, and verify the current delivery mechanism in the AMC's own MCQ specifications before you rely on any assumption about how items are selected. The reported pass standard is a scaled score around 250 on a 0–500 scale, which is your first clue that a raw bank percentage is not the metric the exam uses.
Knowledge is not performance: what a correct answer proves
A correct selected answer proves one thing: on that item, at that moment, with that stem in front of you, you recognised the best option. It does not prove you would have retrieved the same management without five options to choose between, that you would apply Australian rather than home-country practice, or that you could sustain the decision under the fatigue of a three-and-a-half-hour session. Recognition among options is a real skill and the exam rewards it, but it sits at the shallow end of what practising in Australia will demand of you.
The gap matters because banks are optimised to grow the recognition number. You see a stem, choose, read the explanation, and your percentage climbs. What the percentage cannot separate is why you were right — sound reasoning, a half-remembered pattern, or the shape of the options. It also cannot see the three performance skills below, because none of them is scored by a single isolated item answered at your own pace. Training them requires tasks that MCQ drilling does not contain, and measurement that a headline accuracy figure cannot provide.
Skill 1 — Australian guideline localisation
Observable behaviour. Given a common presentation, you default to the Australian standard — the Therapeutic Guidelines (eTG) empirical antibiotic, the Australian Medicines Handbook dose, the National Immunisation Program schedule, the RACGP preventive-care recommendation, the Austroads fitness-to-drive rule — rather than the equivalent from the system you trained in. Localisation failures are quiet: you pick a defensible answer that is simply not the Australian one.
Deliberate-practice task. Build a personal "delta list". Take twenty high-frequency conditions (community-acquired pneumonia, cellulitis, otitis media, asthma escalation, gestational diabetes screening, contraception, warfarin/DOAC choice, childhood immunisation catch-up) and write, side by side, the management you would have given at home and the Australian recommendation with its source. Practise from the Australian column only until it is your first thought.
Feedback source. The primary source is the guideline itself — eTG, the Australian Medicines Handbook, RACGP and the relevant college or public-health authority — not a bank explanation, which may be generic or written for another market. Use bank explanations to find the delta, then confirm against the primary Australian document.
Exit standard. On a fresh, mixed set of localisation-sensitive items you have not seen, you choose the Australian answer at least nineteen times in twenty, and you can name the source for each without looking it up.
Skill 2 — International-graduate transition
Observable behaviour. Beyond drug and dose, you apply the Australian way of practising: shared decision-making and informed consent as the norm, culturally safe care for Aboriginal and Torres Strait Islander patients, the primary-care gatekeeping model, mandatory reporting and notification rules that vary by state and territory, and the ethical and medico-legal expectations that sit behind the Population Health and Ethics strand of the blueprint.
Deliberate-practice task. Take ethics, consent, confidentiality, child-protection and end-of-life items and, before reading the options, write one sentence on the Australian principle at stake. Then choose. This forces you to reason from the framework rather than pattern-match an answer that would pass in another jurisdiction. Add a short weekly block on Aboriginal and Torres Strait Islander health, because it is high-yield and frequently under-practised by overseas-trained candidates.
Feedback source. Australian medico-legal and college resources for the principles; a clinician or peer already working in the Australian system for the judgement calls that guidelines do not resolve cleanly. This is the one skill where a human sense-check materially outperforms both a bank and an AI model.
Exit standard. On unseen ethics and professionalism items you can state the governing Australian principle before you look at the options, and your choice matches it consistently across a mixed block.
Skill 3 — Pacing a long computer-administered session ("adaptive-test pacing")
Observable behaviour. You maintain a steady rhythm of roughly 84 seconds an item — 150 questions across 210 minutes — without banking large time deficits on hard early items or racing the last block. Because the exam is a single computer-administered session and its item-selection mechanism is not something you should assume you can predict, you practise consistent per-item timing rather than a strategy that depends on flagging and freely revisiting questions.
Deliberate-practice task. Do timed blocks of 30–50 unseen items with a visible clock, enforcing a hard per-item ceiling: if you have not chosen by about 100 seconds, commit your best option and move on. Debrief every block on two numbers — how many you got wrong, and how many you got right but slow — because slow-correct items are the ones that sink you at scale.
Feedback source. Your own timing log against your accuracy. The signal you want is stable accuracy at a stable pace, not a high percentage produced by taking three minutes an item.
Exit standard. Across a full-length or half-length unseen block you finish inside time with fewer than five items rushed in the final minutes, and your accuracy in the last quarter is within a few points of your first quarter. Verify the exam's navigation and item-selection rules on the AMC specifications so your practice conditions match the real interface.
A four-week modality ladder
You cannot train these three skills by doing more untimed single items. Climb a ladder from isolated skill to unseen simulation, adding one layer of realism each week.
| Week | Modality | What you do | What it builds |
|---|---|---|---|
| 1 | Isolated skill | Delta lists, principle-first ethics drills, timing ceiling on short blocks — each skill practised alone | Correct Australian defaults; the timing reflex |
| 2 | Coached case | Work through localisation-sensitive and ethics cases with a peer or clinician who challenges your reasoning aloud | Judgement the guideline does not spell out |
| 3 | Timed integrated case | Mixed, timed blocks that combine all disciplines and force localisation, ethics and pace together | Performance under combined load |
| 4 | Unseen simulation | A full-length, unseen, timed mock under test conditions, debriefed by error type | A readiness signal you can trust |
The rungs are deliberate. Isolated practice fixes the default; coaching adds judgement; integration removes the luxury of thinking about one skill at a time; unseen simulation is the only rung that gives you an honest readiness signal, because it is the only one where the questions are new, the clock is real and nobody is helping.
When AI feedback helps, when it misleads, and when you need a clinician
AI feedback is genuinely useful for the mechanical parts of transition: explaining why an option is wrong, generating variations on a stem so you cannot memorise a single phrasing, and surfacing the guideline you should read next. Used that way it accelerates the "find the delta" step of localisation.
It is unreliable exactly where the AMC MCQ is most Australian. A general model may quote the guideline of whichever market dominates its training data, may present a confident but non-Australian dose, and cannot adjudicate a consent or mandatory-reporting question that turns on state legislation. Treat any localisation or medico-legal claim from an AI as a lead to verify against eTG, the Australian Medicines Handbook or the relevant college — never as the source. If you want the discipline for this, our guide on calibrating automated feedback before you trust the score sets out how to check a grader against a known standard.
A clinician or examiner is required for the judgement layer: whether your consent process would satisfy an Australian assessor, whether your management is safe as well as correct, and whether your communication reasoning holds up. No bank percentage and no model output substitutes for that check on the skills the Clinical Examination will later test directly.
A balanced task matrix so you stop practising only what you like
Left to preference, candidates over-practise the strand they already enjoy and starve the rest. Force balance by allocating each week's unseen items against the blueprint, not against your comfort.
| Strand | Target share of weekly unseen items | Localisation-sensitive? |
|---|---|---|
| Adult Health – Medicine | ~30% | High (antibiotics, chronic disease) |
| Adult Health – Surgery | ~20% | Moderate |
| Women's Health – O&G | ~25% | High (screening, contraception, antenatal) |
| Child Health | ~12.5% | High (immunisation, growth) |
| Mental Health | ~12.5% | High (Mental Health Act varies by state) |
| Population Health and Ethics | Woven through all blocks | Very high |
The right-hand column is the point. Localisation risk is not spread evenly; it clusters in prescribing, screening, immunisation and mental-health law. Weight your delta-list work toward those cells and you close the modality gap where it actually costs marks. Our note on why completing a bank is not the same as covering a blueprint explains how to turn this table into a live coverage matrix.
Red flags that you are training recognition, not competence
Five patterns tell you the practice has stopped building anything. Memorised scripts: you recognise the answer from the stem's opening line, not from reasoning. Repeated cases: you are re-seeing the same items, so your percentage measures memory of the bank. Generic feedback: explanations that would read identically for the UK, Canada or Australia — a sign nothing is localising you. Uncalibrated scoring: a rising percentage with no timed, unseen check behind it. No official-rubric check: you never confirm an Australian claim against eTG, the AMH or a college guideline. Any two of these together mean your number is drifting away from your readiness, and it is time to switch from volume to unseen, timed, localisation-weighted measurement.
Frequently asked questions
How do I know whether I have covered the full AMC MCQ blueprint? You do not know it from a completion percentage; you know it from a coverage matrix. List the AMC disciplines and their weightings — Adult Health Medicine and Surgery, Women's Health, Child Health, Mental Health, and Population Health and Ethics woven through — and record, for each, your accuracy on unseen items and whether you have checked its high-yield content against current Australian guidance. A domain is only "covered" when your unseen accuracy is stable and you have localised its prescribing, screening and legal content, not when you have simply attempted questions in it.
Can one question bank be enough for AMC MCQ? One well-constructed bank can carry the knowledge and recognition load for many candidates, particularly if it is large, explained and Australian in its guidance. What one bank cannot give you is an independent readiness signal, because once you have worked through it your score reflects memory of those items. The efficient stack is one primary bank for volume plus a small pool of genuinely unseen, timed questions you keep in reserve to measure transfer — the logic set out in the two-Q-bank rule.
What should I measure instead of my overall Q-bank percentage for AMC MCQ? Measure unseen, timed, mixed accuracy by discipline; your localisation error rate (how often you pick a defensible but non-Australian answer); and your pace stability across a long block. The AMC reports a scaled score, not a raw percentage, so your bank average is not your exam score in the first place — a point we make in full in "Your Q-Bank Percentage Is Not Your Exam Score." Three honest numbers on fresh items beat one flattering number on familiar ones.
When should I stop doing new AMC MCQ questions? Stop adding new volume when your unseen, timed accuracy has plateaued across every blueprint strand and your remaining errors are specific and nameable rather than broad. At that point new questions mostly reassure you; the marginal gain comes from spaced re-testing of your logged misses, localisation drills and full-length timed simulation. Doing more new items past that point buys comfort, not competence.
Which AMC MCQ resource should I use for my weakest component? Match the resource to the type of weakness. If you lack knowledge, use a bank or reference with strong, Australian-referenced explanations. If you lack Australian defaults, use the guidelines themselves — eTG, the AMH, RACGP and college resources — not more generic questions. If you lack pace or transfer, use unseen, timed blocks such as an iatroX baseline. The commonest mistake is answering a localisation or pacing gap with more untimed content practice, which does not touch it.
Editorial notes and references
Written by Dr Kolawole Tytler, NHS GP and founder of iatroX. Last checked 19 July 2026; exam facts are drawn from the AMC's own examination pages and MCQ specifications and should be re-verified against those documents before you sit, as formats and blueprints change. iatroX operates an AMC MCQ question bank, so treat this as an interested but disclosed source; the article confines iatroX to the job the exam itself does not train — independent, unseen measurement of transfer — and does not claim any proprietary selection algorithm. Corrections are welcome via the feedback route on iatrox.com.
References: Australian Medical Council — AMC CAT MCQ examination format and MCQ specifications (amc.org.au); Therapeutic Guidelines (eTG) and the Australian Medicines Handbook for localisation; iatroX AMC MCQ bank (iatrox.com/australia/exam/au-amc); "Your Q-Bank Percentage Is Not Your Exam Score"; "Completion is not coverage"; the AMC MCQ resource decision tree.