If you are relying on a multiple-choice bank to carry your ACEM Fellowship preparation, this is the article to read first, because it names the skills that selected-answer practice cannot assess and gives you a way to train them. Ordinary MCQ practice cannot teach you to construct a structured short-answer response under time, to perform in an interactive OSCE station, or to calibrate your own work against an examiner's rubric. Those three — SAQ construction, OSCE performance and examiner-rubric calibration — are where written-bank scores and real Fellowship results most often diverge. This is the exam-level hub for that single problem; the platform-specific workflows link up to here.
The official format map
Anchor the discussion to what ACEM actually assesses. The Fellowship written examination is two 180-minute papers: a Short Answer Questions (SAQ) paper and a Single Choice Questions (SCQ, multiple-choice only) paper, six hours of written assessment in total. The clinical examination is an OSCE of up to 12 stations, each 11 minutes long — four minutes of reading followed by seven minutes of assessment — for 132 minutes across two consecutive days. Station types span history taking, examination, communication, resuscitation discussion, teaching and case-based discussion. College-reported fees at the last-checked date of 20 July 2026 were AUD $3,145 for the written exam and AUD $4,450 for the OSCE (verify on acem.org.au). The whole assessment is built on the FACEM curriculum and pitched at junior-consultant level. Notice the shape of it: only one of the three assessed papers is pure MCQ. Half the written marks, and all of the clinical marks, live in formats an MCQ bank cannot rehearse.
What a correct selected answer proves — and what it does not
A right MCQ answer proves recognition: shown the correct option among distractors, you can identify it. That is necessary and worth measuring, and it maps well onto the SCQ paper. But recognition is not the same as generation, sequencing or performance. It does not prove you can produce a structured answer from a blank page, prioritise under a station clock, communicate a plan to a distressed relative, or run a resuscitation discussion in real time. The Fellowship is deliberately built to test the gap between recognising the right answer and doing the right thing, observed, at pace. Treating a high MCQ percentage as global readiness is the single most common miscalibration, and it is why your Q-bank percentage is not your exam score.
The practical consequence is that two candidates with identical bank percentages can be miles apart on exam day, because the percentage measures only the recognition modality. The three sections that follow name what that number leaves out, and give each missing skill an observable behaviour and an exit standard, so that a vague intention like "work on my SAQs" becomes something you can actually schedule, do and check.
The three under-tested skills, and how to train each
For each skill, define an observable behaviour, a deliberate-practice task, a feedback source and an exit standard. Vague intentions ("get better at SAQs") do not survive contact with a busy roster; observable behaviours and exit standards do.
SAQ construction
The observable behaviour is a structured written answer that hits the marking points in the marker's order, at the depth the marks demand, within the per-question time budget. The deliberate-practice task is to write full answers to past or bank SAQs under a strict clock, then map what you wrote against a model answer's marking points — not to admire the model, but to find the points you omitted, buried or over-wrote. The feedback source is, ideally, a FACEM who has examined; failing that, a model answer plus a peer using the same rubric, with automated marking (where a bank offers it) as a first-pass drafting signal only. The exit standard is consistent coverage of the marking points at time, across unseen SAQs, without running long on early questions and starving later ones.
OSCE performance
The observable behaviour is a fluent, safe, patient-centred performance across the station types — history, examination, communication, resuscitation discussion, teaching, case-based discussion — inside the four-minutes-reading, seven-minutes-assessment structure. The deliberate-practice task is repeated timed station runs with a role-player and an observer, deliberately varied so you are not rehearsing one script. The feedback source must be a human: a senior or examiner watching the interaction, because tone, safety, prioritisation and rapport are not legible to a text model. The exit standard is stable, safe performance on unseen stations under time, signed off by an experienced observer — not a tally of stations completed.
Examiner-rubric calibration
The observable behaviour is the ability to predict your own mark before you see it, and to be right. The deliberate-practice task is to mark your own SAQ or station against the official-style rubric first, commit to a score, then compare with a senior's mark and analyse the delta. The feedback source is the rubric itself plus an experienced marker. The exit standard is a small, stable gap between your self-mark and the assessor's mark across several unseen tasks — because a candidate who cannot judge their own work cannot triage their own revision, and will either over-prepare comfortable topics or walk in falsely reassured.
A worked example: two candidates, one MCQ score
Consider two FACEM trainees, both sitting a bank at a comfortable 78% overall. On paper they look identical; in the exam they diverge sharply, and the reason is modality.
The first has spread her practice across formats. She writes two timed SAQs a week and maps them against model answers, so her marking-point coverage is climbing; she runs weekly stations with a senior watching, so her communication is fluent and safe under the clock; and she marks her own work against an official-style rubric before every senior review, so her self-mark now sits within a mark or two of theirs. Her 78% is one input among several, all pointing the same way.
The second has spent almost all of his time in the MCQ bank. His recognition is excellent, but he has never written a full SAQ to time, so under pressure he over-writes the first question and starves the last two; he has run only three OSCE stations, all with the same role-player, so his communication is a memorised script that stalls the moment the scenario shifts; and he has never marked his own work, so he cannot tell which of his answers were actually safe. His 78% measured one skill and was silently read as readiness across three.
The number was the same. The preparation was not — and only one of them had trained the modalities the exam actually assesses.
A four-week modality ladder
Do not jump straight to full simulation; climb. The ladder converts an isolated skill into an integrated, unseen performance over roughly four weeks per skill cluster, run in parallel with your ongoing MCQ work rather than instead of it.
- Week 1 — isolated skill. Drill the component alone: SAQ structure on single questions; one OSCE sub-skill (say, structured handover); self-marking against a rubric on short tasks.
- Week 2 — coached case. Add a coach or senior. Write SAQs and run stations with immediate expert feedback, no clock pressure yet, so technique is corrected before it is timed.
- Week 3 — timed integrated case. Introduce the real clock and integrate skills: full SAQ questions at time, full 11-minute stations, self-mark then compare.
- Week 4 — unseen simulation. Sit unseen material under exam conditions — fresh SAQs, unfamiliar stations, a mixed timed SCQ block — and calibrate your predicted marks against the assessor's.
The unseen fourth week is the one candidates skip, and it is the only week that measures readiness rather than rehearsal.
When AI feedback helps, when it misleads, and when a human is required
Automated feedback has a real but bounded role, and it is worth being precise about the boundary. AI marking is useful as a fast first pass on SAQ drafts — flagging omitted marking points, structural sprawl or an unaddressed part of the stem — and for generating unseen transfer questions on a knowledge point. It is unreliable as a final SAQ mark, because it can reward fluent, well-structured prose that is subtly wrong, and it cannot see the safety and prioritisation judgements a rubric rewards. And it is simply the wrong tool for the OSCE: interactive performance, tone, rapport and real-time safety require a human examiner. Before you trust any automated score, calibrate it against a human-marked sample; an uncalibrated automated mark is a number, not a judgement. This is also where iatroX sits honestly in the picture: iatroX covers ACEM Primary, so it is the layer that measures the basic-science foundation beneath these Fellowship skills on unseen MCQs — it is not a Fellowship SAQ marker or an OSCE simulator, and should not be used as one.
A balanced case and task matrix
Candidates gravitate to scenarios they already handle well, which quietly narrows their practice. Prevent this with a simple matrix: list the station types down one axis and the acuity or population (resuscitation, adult undifferentiated, paediatric, mental health, communication-heavy, procedural) across the other, and insist on covering the cells you avoid. The same discipline applies to SAQs: rotate systems and question stems so you are not repeatedly writing the cardiology answer you already know. A matrix turns "practise more cases" into a countable coverage task and stops comfortable repetition from masquerading as breadth.
| Station type | Resuscitation | Adult undifferentiated | Paediatric | Mental health | Communication-heavy | Procedural |
|---|---|---|---|---|---|---|
| History | ||||||
| Examination | ||||||
| Communication | ||||||
| Resuscitation discussion | ||||||
| Teaching | ||||||
| Case-based discussion |
Tick a cell only once you have run that combination as an unseen, timed task with feedback. If a whole row or column is empty, that is your next fortnight, not the topic you enjoy.
Reading your readiness signals honestly
Three signals tell you the modality work is landing. First, your predicted marks converge on your assessors' marks — the calibration gap shrinks, which means you can finally triage your own revision instead of guessing. Second, your SAQ marking-point coverage rises without your total time rising, which means structure and pacing have become automatic rather than effortful. Third, your OSCE performance holds on stations you have never seen, watched by someone new, which means you are performing a skill rather than reciting a rehearsed case. If instead your MCQ percentage is the only number moving, that is the warning sign: you are polishing the one modality that was already strong and leaving the two that decide the result untrained.
Red flags that you are training the wrong thing
- Memorised scripts. If your OSCE communication sounds identical regardless of the patient, you are performing recall, not rapport; examiners notice.
- Repeated cases. Re-running the same three stations inflates confidence without building adaptability; unseen material is the test.
- Generic feedback. "Good, work on timing" changes nothing. Demand feedback tied to specific marking points or observable behaviours.
- Uncalibrated scoring. A mark from a rubric no one has anchored to the official standard is noise. Calibrate against a human marker first.
- No official-rubric check. If you have never marked your work against official-style ACEM criteria, your self-assessment is untested — and so is your readiness.
The bottom line
An MCQ bank is a genuine asset for the ACEM Fellowship — it builds and measures the recognition that the SCQ paper rewards and that much of the SAQ paper assumes. What it cannot do is generate a structured written answer from a blank page, carry you through a live station, or judge your work against a rubric. Those are separate skills, with separate feedback sources, and they are where the result is won and lost. Keep your bank running for what it does well, keep an unseen foundation-measurement layer such as iatroX beneath it at ACEM Primary level, and spend the time the bank frees up on the three modalities it cannot touch. Train the format you are sitting, not the format that is easiest to score.
Frequently asked questions
How do I know whether I have covered the full ACEM Fellowship blueprint? You audit coverage against the FACEM curriculum and the official assessment structure, not against a bank's completion bar. Build a blueprint-coverage matrix of domains against your evidence — unseen first-attempt accuracy for knowledge, observed sign-off for performance — and treat any empty or weak cell as uncovered regardless of how many questions you have answered. Completion is not coverage; a finished bank can still leave whole clinical areas untested.
Can one question bank be enough for ACEM Fellowship? No single MCQ bank can be sufficient, because the Fellowship is not a single-format exam. Even an excellent SCQ bank leaves the SAQ paper's construction demands and the entire OSCE untrained. A good written bank can be your engine for the SCQ paper and part of the SAQ paper, but it must be paired with structured SAQ practice, observed OSCE work and calibration against official material.
What should I measure instead of my overall Q-bank percentage for ACEM Fellowship? Measure unseen, timed first-attempt accuracy by domain (not cumulative percentage); your pace against the per-item and per-station budgets; SAQ marking-point coverage at time; the gap between your predicted and actual marks; and retention on delayed retests. These track readiness; an overall percentage tracks mostly how much of the bank you have seen.
When should I stop doing new ACEM Fellowship questions? Stop adding new items when your coverage floors are met, your unseen timed performance is stable across at least two sessions, your pace holds, corrected rules survive delayed retests, and your standard is calibrated against official material. Past that point, extra questions buy reassurance rather than readiness, and your remaining time is better spent on SAQ and OSCE performance.
Which ACEM Fellowship resource should I use for my weakest component? Match the tool to the deficit. If SCQ knowledge is weak, use a Fellowship MCQ bank and, for the underlying basic sciences, an unseen Primary-level measurement layer such as iatroX. If SAQ construction is weak, use timed written practice with model answers and, ideally, examiner feedback. If OSCE performance is weak, nothing but observed, timed, interactive practice with a senior will do — no bank substitutes for it.
Editorial notes and references
Written by Dr Kolawole Tytler, NHS GP and founder of iatroX. Last checked 20 July 2026. Exam format and college-reported fees are drawn from ACEM's own pages and were current at that date; verify on acem.org.au. Disclosure: iatroX operates an ACEM Primary / foundation-level question bank; this article confines iatroX's role to unseen-MCQ measurement of the underlying knowledge and states plainly that it is not a Fellowship SAQ marker or OSCE simulator. Corrections are welcome via the feedback route on iatrox.com. References: ACEM Fellowship examinations and FACEM curriculum (acem.org.au); Your Q-Bank Percentage Is Not Your Exam Score; Question-bank completion is not coverage; Calibrating AI-graded SAQs and OSCEs.
Complete a fresh ACEM Primary-level baseline in iatroX, then pair it with SAQ and OSCE practice →