If you are preparing for the Key Feature Problem exam using advice about writing one answer per line, avoiding undercoding, and never stacking two answers together, you are preparing for an exam that no longer exists. From the 2025.2 sitting, the KFP changed substantially: it is now 70 independent clinical scenario stems, each answered by selecting multiple responses from a list, on computer-readable answer sheets. There is no handwritten short-answer component. A great deal of the KFP folklore, and a great deal of the material still circulating, describes the previous exam. Confirm the current format against the RACGP's own guidance before you build a study plan around anything.
Key takeaways
- The KFP is now 70 independent scenario stems, answered by multiple selection rather than free text.
- Each item has between two and six correct responses, chosen from a longer list of options.
- Each question carries equal weight, at roughly 1.43 per cent of the total.
- The old rules about undercoding, specificity of wording and stacking answers on a line no longer apply.
- What has not changed is the underlying skill: identifying the key features that actually decide the case.
What actually changed
The old KFP asked you to write. You were given a case, asked for a specific number of answers, and you typed or wrote them, one per line. That format generated the exam's distinctive folklore: answers had to be specific enough to score but not padded, you were penalised for insufficient detail, and putting two answers on one line could cost you both.
The new KFP asks you to select. You are given a scenario and a list of options, and you choose the responses that are correct, with between two and six correct answers per item. The answer is shaded on a computer-readable sheet. There is no free text.
Seventy stems, each independent, each worth the same, at approximately 1.43 per cent of the paper.
That is a different exam, and it is a different skill.
What this means for the advice you have been reading
Be ruthless about discarding what no longer applies, because the old material is everywhere and much of it is not labelled as historical.
Undercoding is gone. You are no longer being marked on whether you wrote "type 2 diabetes secondary to metabolic syndrome" rather than "diabetes." You are selecting from options that have already been worded for you.
The one-answer-per-line rule is gone. There are no lines.
The specificity-of-wording anxiety is gone. The examiners have written the options, so you cannot be penalised for imprecise phrasing.
The exam is no longer handwritten. Which removes a category of preparation, and a category of anxiety, entirely.
If a course, a book or a forum post is teaching you these things, it is teaching you a real exam that you will not be sitting.
What has not changed at all
Here is the part that matters, and it is the reason the exam is still called the Key Feature Problem.
The underlying construct is identical. A key feature is the specific decision or action that most affects the outcome for that patient: the thing that, if you get it wrong, the case goes wrong. The exam is still testing whether you can identify those decisions, in a realistic general practice scenario, under time pressure.
So the central skill you must build is unchanged. For every clinical scenario you meet in your preparation, ask the question the exam is asking: what is the thing here that actually matters? What must not be missed? What must be done first? What would a competent, independent Australian GP do that a less safe one would not?
The format changed. The competency did not.
The new risk: selecting too many
The old format punished writing too little. The new format introduces a different hazard, and it is worth thinking about deliberately.
When an item asks you to choose from a long list, and several options look reasonable, the temptation is to hedge by selecting broadly. That instinct comes from a good clinical place, because in practice you would rather consider too much than too little.
The exam is not practice. It is asking which responses are correct, and the number of correct responses is bounded, between two and six. Selecting options that are not correct does not demonstrate thoroughness. It demonstrates that you cannot discriminate, which is precisely what the item was constructed to test.
So resist the urge to shade everything plausible. Decide what is actually correct, select that, and stop.
Prioritisation is still the whole game
General practice is a specialty of triage and of doing the right things in the right order with limited information, and the KFP has always been built around that.
The scenarios will present you with a patient about whom you know less than you would like, and they will ask what you would do. The options will contain actions that are reasonable, actions that are unsafe, actions that are irrelevant, and actions that are correct but premature.
The reasoning that separates them is the reasoning of an independent GP: what is the most likely and most dangerous possibility here, what will change my management, what can wait, and what would be unsafe to omit?
That is trainable, and it is trained by practising cases and interrogating your decisions, not by memorising content.
Prepare with the AKT, not separately
A practical planning point.
The AKT and the KFP test different capabilities, but their clinical content overlaps almost completely, and preparing for them in sequence duplicates effort.
Prepare the content once, across the breadth of Australian general practice, and then practise it in two modes: as knowledge questions for the AKT, and as decision scenarios for the KFP. The knowledge you build for one strengthens the other, and the case practice you do for the KFP deepens the clinical understanding that the AKT tests.
Candidates who treat them as two separate projects spend more time and learn less.
Use current Australian guidance
The final and unglamorous point.
KFP cases are developed in line with current guidelines, and the guidelines that matter are the Australian ones. Australian antibiotic guidance, Australian preventive activity recommendations, Australian cardiovascular risk assessment and Australian prescribing conventions are what is being examined.
A candidate who has built their reflexes on another country's guidance can reason impeccably to an answer that is not the answer, and it is a calibration problem rather than a knowledge problem. It needs to be corrected deliberately.
Where iatroX fits
iatroX's RACGP bank supports the decision reasoning the KFP tests, with an adaptive engine that targets the areas where your prioritisation is genuinely weak rather than the topics you enjoy, and spaced repetition to hold the breadth of general practice across a long preparation. Missed questions can be opened in the Socratic Tutor, which asks you what must happen first and why before it explains, which is precisely the reasoning step a key feature question is built around. Try it with free sample questions at iatroX. For the knowledge paper that shares its content, see the RACGP AKT and Australian calibration.
Frequently asked questions
What is the current KFP format? From the 2025.2 sitting, 70 independent clinical scenario stems, each answered by selecting multiple responses from a list, with between two and six correct answers per item, on computer-readable answer sheets. There is no handwritten short-answer component.
Do the old rules about undercoding still apply? No. The rules about writing sufficiently specific answers, one answer per line, and not stacking responses were features of the previous free-text format and no longer apply, because the examiners now write the options for you.
What is the new risk in the KFP? Over-selection. When several options look plausible, the temptation is to hedge and select broadly. The number of correct responses is bounded, and selecting incorrect options demonstrates poor discrimination rather than thoroughness.
Should I prepare for the AKT and KFP together? Yes. They test different capabilities but their clinical content overlaps almost completely. Build the content once, then practise it in two modes: as knowledge questions and as decision scenarios.