What the ACEM Fellowship SAQ Paper Actually Tests

By Dr Liam Stephens, FACEM. Based on all 296 scored short-answer questions listed in the ACEM Fellowship Written Examination Reports for 2021.1 to 2026.1. Last updated September 2026.

Every ACEM written examination report lists the presentation or scenario behind each SAQ, the marks available, and in an appendix, the average mark the cohort scored on it. Put 11 reports together and you have 296 questions with a difficulty rating attached to each. That is enough to say, with numbers rather than folklore, what the paper tests, how heavily, and where candidates actually drop marks.

How the paper is built

The SAQ paper is three hours. From 2021 to 2025 it contained 26 to 28 questions worth 360 marks; questions are worth 12 or 18 marks. The 2026.1 paper broke the pattern with 26 questions and 330 marks. Early in the series (2021) there were eight 18-mark questions per paper; from 2022.2 there have usually been four.

Because marks are the currency, not questions, the arithmetic is worth doing once. Three hundred and sixty marks in 180 minutes is two marks a minute. A 12-mark question earns six minutes and an 18-mark question nine. Candidates who give every question the same time are over-investing in the 12-markers and rushing the ones that carry half as much again.

The 18-mark questions are not harder. Across the series the cohort scored 67.3% on 18-mark questions and 67.8% on 12-mark questions. They are simply longer, and usually anchored in a resuscitation, trauma, obstetric or paediatric emergency. There is no reason to fear them and every reason to give them their nine minutes.

A quarter of the paper is paediatric, every time

The reports state the number of paediatric SAQs in each paper. It has never been below five or above nine, and the share of marks has ranged from 18% (2022.2) to 33% (2025.2), averaging 26%. This is the single most reliable structural fact about the paper. A candidate who is weak in paediatrics is weak in a quarter of the exam.

It is also worth knowing that paediatric questions are not scored worse than adult ones. Where a question can be identified as paediatric from its title, the cohort averaged 68% on it, against 67% for the rest. The topic is large, not unusually hard.

Where the marks are: 21 categories

We coded every question into one of 21 categories from its scenario title. Some calls are judgement (a "chest pain in a rural location" question is coded as cardiology; a "paediatric NAI" question as non-clinical because the marks are about process). The table shows the share of total marks, how many of the 11 papers the category appeared in, and the average score candidates achieved.

The near-certainties

Seven categories appeared in at least 10 of the 11 papers: cardiology and ECG, toxicology, obstetrics and gynaecology, and general medicine in all 11; trauma, resuscitation and neurology in 10. Between them these seven carry just over half the marks in the series. Nothing about that is surprising, but the reports let you go one step further and ask which of them candidates handle well.

• Obstetrics and gynaecology is the best-scored category in the exam at 75%. Premature labour, precipitous delivery, postpartum haemorrhage, pre-eclampsia, early pregnancy bleeding and ovarian torsion have all appeared, and candidates answer them well. These are marks you should expect to collect.
• Trauma and resuscitation score around 68% to 70%, in line with the paper average, with a wide spread between questions. Abdominal trauma (2026.1, 80%), chest trauma (2026.1, 83%) and geriatric chest injury (2024.2, 79%) were among the easiest questions in the series. Spine trauma (2022.1, 38%) was the hardest question in all 11 papers.
• Neurology is the lowest-scoring of the big categories at 64%. Generalised weakness (47%), headache (53%), motor weakness (58%), neuro assessment (62%), ataxia (58%) and dizziness (64%) were all below average. Neurology questions tend to ask for a differential and a structured assessment rather than a protocol, and candidates find that harder to score on.
• Toxicology appears in every paper and scores below average (66%). The named-agent questions (snake bite 83%, beta-blocker overdose 73%, toxic alcohols 62%) are answered better than the questions with generic titles: “Medication” (2022.1) and “Toxicology” (2024.2) scored 45% and 46%, the second and third lowest in the whole series.

The non-clinical category is bigger than most candidates think

Questions on administration, ethics, quality, evidence-based medicine and systems appeared in nine of the 11 papers and carried 7.6% of all marks, more than toxicology. The topics were: an impaired clinician (twice), change management, root cause analysis, patient flow and access block, a vulnerable patient (twice), critical appraisal, medico-legal frameworks, end of life care (twice), key performance indicators, mentoring, health promotion, restrictive practices, mandatory reporting, cognitive bias and medical error, literature appraisal, triage systems, intimate partner violence, non-accidental injury, research methodology, frequent presenters, and discharge against medical advice.

Scores in this category are polarised. End of life care (88%, 2024.2) and paediatric NAI (86%) were among the easiest questions in the whole series. Root cause analysis (47%), literature appraisal (54%), cognitive bias (55%), health promotion (58%) and mentoring (59%) were among the hardest. The pattern is that candidates do well when the non-clinical question is about a patient in front of them and poorly when it asks them to describe a process or framework in the abstract. The fix is unglamorous: know the named frameworks (RCA, the steps of critical appraisal, the components of a triage system, the levels of the hierarchy of evidence) well enough to write them down as a list. Our admin, medicolegal and ethics guide and statistics and EBM guide cover the material.

The small categories that always turn up

Ophthalmology has appeared in nine of the 11 papers, always as a single 12-mark question: visual disturbance, loss of vision (three times), red eye, orbital cellulitis, anterior uveitis and a generic ophthalmology question. Candidates average 69%, but the 2021.1 visual disturbance question scored 52%. One question is 3% of the paper; a category that turns up nearly every sitting for 3% of the marks is worth an afternoon.

The same logic applies to ENT and dental (seven papers, including dental pain, peritonsillar abscess, epistaxis and foreign bodies), renal and urology (eight papers; urology questions scored 53% in both 2021.1 and 2026.1), and environmental medicine (five papers; diving medicine 52%, electrical injury 55%, hypothermia twice, hyperthermia). None of these is large. All of them are predictable.

The 20 hardest questions in the series

These are the questions with the lowest cohort average as a percentage of the marks available. Every one was answered, on average, at or below the pass mark.

Read down the list and three types of question recur.

1. The second-order clinical topic. Spine trauma, a mediastinal mass, SIADH, diving medicine, generalised weakness, a neonatal lesion, liver disease. These are conditions every registrar has met, but not ones most candidates have revised to the depth of writing a structured 12-mark answer. They are the topics that live in the margins of the curriculum and get skipped in a time-pressured study plan.
2. The interpretation question. An arterial line trace (51%), a retrograde urethrogram (51%), a VBG (55%, 2024.2), an invasive monitoring device (58%). The reports show that questions built around interpreting a trace, image or result score about three points below the paper average. Practise writing down what you see in a structured way, not just recognising it.
3. The process question. Root cause analysis, literature appraisal, cognitive bias, mentoring. As above: candidates lose marks on frameworks they cannot reproduce as a list.

What is not on the list is instructive too. Apart from a “challenging intubation” question in 2021.2, there is no cardiac arrest, airway, major haemorrhage, anaphylaxis or sepsis question among the hardest 20. The core resuscitation curriculum is well learned. Marks are lost at the edges.

The easiest questions

End of life care, a paediatric rash, paediatric NAI, paediatric syncope, paediatric asthma, cardiac arrest, pulmonary embolism, trauma resuscitation, precipitous labour. These are high-frequency ED presentations with clear management frameworks. When the paper asks about something you do every week, candidates score 80% or more. The distance between those scores and the 45% to 55% on the hardest questions is where the SAQ paper is decided.

Topics that came back

Individual scenarios repeat more often than most candidates expect. Within these 11 papers:

• Syncope appeared in 2023.1, 2024.1 (paediatric), twice in 2024.2 (both ECG-based) and 2026.1.
• Loss of vision or visual disturbance in 2021.1, 2021.2, 2022.1, 2025.1 and 2025.2.
• Premature or preterm labour in 2023.2, 2024.1 and 2024.2; precipitous labour in 2021.2; pre-eclampsia in 2024.2 and 2025.2.
• Pain management, sedation and analgesia in 2021.1, 2021.2, 2022.1 (twice), 2022.2 and 2023.1.
• Hand injury in 2022.1, 2022.2 and 2023.1.
• Guillain-Barré syndrome in 2021.2 and 2024.1. Subarachnoid haemorrhage in 2024.1 and 2025.2. Hypothermia in 2022.1 and 2024.1. Chickenpox in 2021.2 and 2025.2. Iron overdose in 2021.2 and 2025.2. Retrograde urethrogram in 2021.2 and 2023.1. VBG interpretation in 2021.2 and 2024.2. End of life care in 2022.2 and 2024.2. An impaired clinician in 2021.1 and 2024.1. Non-accidental injury in 2023.1 and 2024.2. Neonatal sepsis in 2023.2 and 2026.1, and neonatal resuscitation in the same two papers.

A scenario returning after two to four sittings is normal. If a topic on this list has not appeared for two years, it is not "done".

Using this to plan revision

A study plan built from the curriculum treats every domain as equally likely. A study plan built from 11 real papers does not. In priority order:

1. Paediatrics across every system. A quarter of the marks. Resuscitation, sepsis, the crying or cyanotic infant, seizures, asthma and croup, intussusception and abdominal pain, NAI, toxic ingestion, syncope and cardiac presentations have all appeared.
2. The seven near-certain categories, with extra time on the ones that score badly: neurology (structured assessment and differentials, not just stroke) and the generic toxicology approach to an unknown ingestion.
3. The non-clinical frameworks, learned as reproducible lists. This is 8% of the marks and it is where the most avoidable losses are.
4. Interpretation practice: ECGs, blood gases, X-rays, ultrasound findings and monitoring traces, written up in a structured format under time pressure.
5. The one-question categories: ophthalmology, ENT and dental, urology, environmental. An afternoon each buys 3% of the paper.
6. The second-order topics from the hardest-questions list. Spinal injury, liver disease, endocrine emergencies beyond DKA, haematological and oncological emergencies.

Whatever you cover, practise it as a written answer against a marking scheme and against the clock. The reports show that the average candidate on the SAQ finishes within one question's worth of marks of the pass line; our guide to how marks are lost in the SAQ explains where those marks go.

Appendix: every SAQ scenario, 2021.1 to 2026.1

The full list from the reports, with the marks available and the cohort average for each question. Questions excluded from the results after review are omitted. The 2022.2 hip dislocation question was reduced to seven marks after part of it was removed.

Sources and method

Question titles, marks and cohort means are transcribed from the "Written Examination Content" section and Appendix 4 of each ACEM Fellowship Written Examination Report, 2021.1 to 2026.1. Category coding is by ACEMCQ from the scenario titles alone; the reports do not publish the questions themselves, so a title such as "Medication" or "Risk stratification" is coded on its face. Paediatric identification uses the report's own counts for share of marks and the question title for per-question scoring. Percentages are cohort mean divided by marks available. ACEMCQ is independent of ACEM and this analysis is not endorsed by the College.

Frequently asked questions

How many SAQs are in the ACEM fellowship written exam?

Between 26 and 28 questions in a three-hour paper, worth 360 marks in total (330 in 2026.1). Most questions are worth 12 marks; three to eight per paper are worth 18.

What proportion of the ACEM SAQ paper is paediatrics?

Between 18% and 33% of the marks in every sitting from 2021 to 2026, averaging 26%. That is five to nine questions per paper.

Which topics come up most often in ACEM SAQs?

Trauma (10% of marks), cardiology and ECG (9%), resuscitation and airway (8%), non-clinical topics such as administration, ethics and evidence-based medicine (8%) and toxicology (7%). Cardiology, toxicology, obstetrics and gynaecology appeared in all 11 papers; trauma, resuscitation and neurology in 10.

Which ACEM SAQ topics do candidates score worst on?

By category, orthopaedics and wounds (64%), neurology (64%), environmental medicine (62%), investigation interpretation (64%) and endocrine (65%). The lowest-scoring individual questions were spine trauma (38%), a medication question (45%), a toxicology question (46%), root cause analysis (47%) and generalised weakness (47%).