Read the honest finding first. As of 19 July 2026, the public pages for PassAMC (passamc.org) and PassAMCQ (passamcq.com) describe human-written AMC MCQ question banks with detailed explanations and progress analytics — not a conversational AI tutor. So if you came here expecting to interrogate a built-in AI, adjust your plan: this workflow gives you a reusable prompt-and-verification discipline to apply to whatever explanation or general-purpose AI you use alongside these banks, so every missed AMC MCQ item builds reasoning rather than recognition. It addresses the written MCQ only.
What PassAMC and PassAMCQ offer for AMC MCQ right now
Figures are vendor-reported and were last checked on 19 July 2026. Confirm current counts and prices on each site.
| Item | PassAMCQ (passamcq.com) | PassAMC (passamc.org) |
|---|---|---|
| Question volume | "1,600+" to "1,800+ and growing" high-yield AMC MCQ questions | "Over 2000 AMC past paper MCQs" |
| Mock exams | Two full mocks, 100 questions each | Timed exam mode |
| Price | 1 month A$150; 3 months A$250; 6 months A$350; 12 months A$450 (vendor-reported) | Not displayed until checkout |
| Explanations | Detailed explanations; "smart learning tools" for weakness tracking | Human-authored explanations "written by qualified doctors" |
| AI tutor | Not advertised | Not advertised; vendor emphasises human expertise over AI |
| Components | AMC MCQ only | AMC MCQ only |
The takeaway: both are legitimate, Australia-focused MCQ banks with human explanations and analytics, and either is worth including in a revision stack. Neither, on the evidence of its public pages, is an "AI tutor" product. That is not a criticism — human-written explanations are often more trustworthy — but it means the "AI answers" step of any prompt workflow refers to a tool you add, such as iatroX's citation-first assistant, not something inside PassAMC or PassAMCQ. Treat that as the ground truth for the rest of this article.
The exam you are actually sitting
The AMC MCQ is named the Computer Adaptive Test (CAT) MCQ. Per the AMC, it is 150 single-best-answer questions, one correct of five, all scored, in one 3.5-hour session, computer-administered at Pearson VUE. The AMC's public page does not describe a difficulty-adjusting algorithm, so prepare for a fixed-length 150-item paper. It covers the disease process, clinical examination and diagnosis, investigation, and therapy and management, and it gates the AMC Clinical Examination. Because it tests Australian practice, the correct authorities are Therapeutic Guidelines (eTG), the Australian Medicines Handbook, the National Immunisation Program schedule and RACGP guidance — not an overseas formulary and not US board norms.
Prompt taxonomy: six jobs a good prompt does
Whether you are testing your own recall, pushing a bank's written explanation, or querying an external AI, a prompt should do one of six jobs: probe the mechanism; surface the discriminating features between look-alike answers; force option elimination with reasons; run a guideline verification against the correct jurisdiction and date; test a counterfactual ("what would change the answer?"); or drive retrieval testing on a fresh case. The point of every prompt is to make you generate reasoning before you receive it.
Twenty-five reusable prompts, grouped by the six common error types
Use these against explanations or an external AI. They contain no proprietary question text; fill the brackets with your own case.
1. Knowledge gap / mechanism
- "Explain, in three steps, the pathophysiological mechanism that makes [condition] present with [feature]."
- "What is the single most important mechanism linking [drug] to [adverse effect]?"
- "Give the one-sentence rule that explains why [investigation] is first-line for [presentation]."
- "Summarise the natural history of [condition] and the point at which management changes."
2. Misread stem / discriminating features
- "List the exact words in a stem that would change the answer from [A] to [B]."
- "What single discriminating feature separates [diagnosis 1] from [diagnosis 2] in this presentation?"
- "Rewrite this lead-in as 'most likely', 'next best test' and 'definitive test' and give the answer for each."
- "Which one demographic or timing detail most changes the differential here, and why?"
3. Premature closure / option elimination
- "Give a specific reason to reject each of the other four options, not just why the key is right."
- "What is the strongest alternative diagnosis, and what finding would make it the answer?"
- "Steel-man the option I chose: under what circumstances would it be correct?"
- "List the red-flag features that must be excluded before accepting this diagnosis."
4. Guideline / jurisdiction error
- "State whether this management reflects Australian practice, citing Therapeutic Guidelines or an RACGP source and the date."
- "What is the first-line empirical antibiotic for [infection] per Australian Therapeutic Guidelines, and how does it differ from UK or US practice?"
- "Does the vaccine advice here match the current National Immunisation Program schedule? Quote the relevant point."
- "Is this drug and dose consistent with the Australian Medicines Handbook, and is it PBS-relevant?"
5. Calculation error
- "Work this dose calculation step by step, stating units at each line, then give the final answer."
- "Recompute this weight-based paediatric dose and flag any unit conversion I might have missed."
- "Show the fluid-rate calculation and the assumption behind each number."
- "Estimate the pre-test probability, then apply the likelihood ratio and state the post-test probability."
6. Time-pressure / retrieval testing
- "Give me one fresh single-best-answer question on the same concept with five plausible options, but do not reveal the answer."
- "Quiz me on three variations of this concept in different clinical contexts; withhold answers until I respond."
- "Turn this fact into a 20-second recall cue I can self-test in a mixed block."
- "Write a harder version of this item where the obvious answer is a distractor."
- "Ask me the single question that would prove I understand this rather than remember it."
The anti-answer-leak rule
Before you open any explanation or send any prompt, commit your answer and a one-line rationale in writing. If you read the explanation first, you rob yourself of the retrieval attempt and mistake recognition for knowledge. Commit, then verify. This single discipline is what separates a study session that raises your score from one that merely feels productive.
The jurisdiction check
Make one reusable prompt non-negotiable for every management, prescribing or public-health item: "State whether this answer reflects the Australian (AMC) jurisdiction, name the specific guideline (Therapeutic Guidelines, RACGP, the National Immunisation Program or the Australian Medicines Handbook), and give the date of that guidance." An answer that cannot name an Australian source and a date is not yet safe to bank. This is the highest-yield habit for IMGs, because most freely available explanations and general AI models default to US or UK norms.
Create a misconception record
For every genuine error, write four lines: the incorrect rule you were using; the corrected rule in one sentence; one transfer question that tests the corrected rule in a new context; and a review date. A running misconception log is worth more than any highlight, because it converts scattered mistakes into a short, personal, high-yield revision list you can re-test in the final week.
Weekly verification sample
Once a week, take a small random sample — say 8 to 10 — of the explanations or AI outputs you relied on and independently check them against official or primary sources: the AMC materials, eTG, the Australian Medicines Handbook, RACGP guidance. Log any discrepancy. This matters doubly when the "answer" came from a general AI, which can hallucinate a plausible but non-Australian recommendation. If you want a structured method, our guide to auditing an AI exam tutor sets out grounding, answer-leakage and hallucination checks, and calibrating automated feedback covers when to trust a score.
A worked seven-day plan
An IMG six weeks out, using PassAMC or PassAMCQ for question volume and human explanations, and iatroX's AMC CAT MCQ bank and its Socratic Tutor for unseen transfer questions and citation-first verification. No proprietary-algorithm claims are made about any tool.
| Day | PassAMC / PassAMCQ (one job: volume + explanations) | iatroX (one job: unseen transfer + verified AI check) |
|---|---|---|
| Mon | 40 Qs; commit answers first; write misconception records | — |
| Tue | Re-read only the explanations behind Monday's errors | Run the jurisdiction prompt on each error; 12 unseen Qs |
| Wed | 40 Qs; tag every error by type | — |
| Thu | Mock 1 (100 Qs) in timed mode; tally error taxonomy | Verify 8 tutor/explanation outputs vs eTG/AMH; log gaps |
| Fri | Review mock; update misconception log | 12 unseen transfer Qs on the week's weak concepts |
| Sat | Light review of the misconception log | — |
| Sun | Rest | 15 unseen mixed Qs; compare week-on-week trend |
Decision checklist: continue, supplement, switch or stop
| Situation (measurable) | Action |
|---|---|
| Still building volume; misconception log still growing fast | Continue the bank as primary driver |
| Explanations thin on Australian sourcing; unseen accuracy flat | Supplement with a citation-first AI and Australian primary sources |
| Repeated guideline errors trace to non-Australian explanations | Switch the offending source; re-verify against eTG/RACGP |
| Coverage, unseen accuracy, pacing and retention at target | Stop grinding; move to full mocks and maintenance |
Bottom line
PassAMC and PassAMCQ are credible, Australia-focused MCQ banks with human-written explanations, and either is worth a place in the stack — but neither is an AI tutor, so the "AI answers" in any prompt workflow come from a tool you add. Commit before you verify, force a jurisdiction check on every management item, keep a misconception log, and sample-audit your AI outputs weekly against Australian primary sources.
Frequently asked questions
Is PassAMC and PassAMCQ enough for AMC MCQ on its own? Either can carry question volume and offers human-written explanations, which many candidates find sufficient for content. But "enough" also needs an unseen readiness signal and verified Australian sourcing, and once you have worked through a bank your score reflects recognition as much as knowledge. Pair the bank with fresh, timed questions and check management answers against Australian guidance before deciding you are ready.
Which AMC MCQ component does PassAMC and PassAMCQ not reproduce well? Both are written-MCQ products and do not stand in for the AMC Clinical Examination. Equally important, neither advertises an AI tutor as of 19 July 2026, so if your study plan relies on interrogating an AI you must add that tool separately — do not assume a conversational tutor is built in. Their strength is human explanation and volume, not interactive reasoning.
How should I verify PassAMC and PassAMCQ AI answers for AMC MCQ? Because the banks are human-written, the "AI answers" you need to verify usually come from an external assistant you bolt on. Commit your own answer first, then require the AI to name an Australian source and date via the jurisdiction prompt, and once a week independently check a random sample of outputs against the AMC materials, Therapeutic Guidelines, the Australian Medicines Handbook and RACGP guidance, logging any discrepancy. Never accept an unsourced management recommendation.
When should I stop using PassAMC and PassAMCQ and move to mixed mocks? Move to predominantly mixed, timed mocks when every domain has cleared a coverage floor, your first-attempt accuracy on unseen blocks is stable across two weeks, and you are within about four weeks of the exam. Use the banks' full mocks in timed mode for full-length rehearsal at the 150-item, 3.5-hour load, and keep topic practice only for lagging areas.
How should I combine PassAMC and PassAMCQ with iatroX without duplicating practice? Give each a single job. Use PassAMC or PassAMCQ for volume and explanations; use iatroX for small, unseen, timed transfer blocks and, through its citation-first Socratic Tutor, for the verified AI check the banks do not provide. Never review the iatroX items one by one, or you lose the unseen signal. The two-Q-bank rule explains how to avoid overlap.
Editorial notes and references
Written by Dr Kolawole Tytler, NHS GP and founder of iatroX. Last checked 19 July 2026. Vendor figures (PassAMCQ and PassAMC question counts, mock counts and prices) are vendor-reported as at that date and can change; verify current counts and prices on each site. Honesty note: neither PassAMC (passamc.org) nor PassAMCQ (passamcq.com) advertised a conversational AI tutor on its public pages as at 19 July 2026, so this workflow treats "AI answers" as coming from an external, citation-first assistant, not a built-in feature. Disclosure: iatroX operates a competing AMC CAT MCQ question bank and a Socratic Tutor; here iatroX is confined to the unseen-measurement and verified-AI roles these banks do not claim, and is not positioned as a clinical-examination product. Corrections are welcome via the feedback route on iatrox.com.
References: AMC — Computer Adaptive Test (CAT) MCQ Examination and MCQ specifications (amc.org.au); AMC via Pearson VUE (pearsonvue.com/amc); PassAMCQ (passamcq.com) and PassAMC (passamc.org) product pages; Therapeutic Guidelines (tg.org.au), Australian Medicines Handbook (amhonline.amh.net.au) and the National Immunisation Program schedule (health.gov.au); and the iatroX comparison hub and AMC CAT MCQ bank.
Open a fresh AMC MCQ item in iatroX and run the verification prompt →