Most registrars know the KFP has "partial credit" and an "over-selection penalty." But when you ask how those mechanisms actually work, or how they interact, the answers tend to be vague. That vagueness costs marks.
This article breaks down the KFP scoring model using verified data from the RACGP 2026.1 KFP Public Exam Report and the official RACGP Examination Guide. If you understand how the scoring works before you sit the exam, you can make deliberate decisions about what to select, what to leave out, and when to stop.
The KFP is not a multiple-choice exam
This distinction matters more than most registrars appreciate. The AKT uses single best answer (SBA) questions: one correct answer from a list, no penalty for guessing, one mark per question. It is a classic MCQ format.
The KFP uses a fundamentally different approach. Since the 2025.2 sitting, the KFP is delivered as a paper-based exam consisting of approximately 70 standalone clinical scenarios, each paired with a single multiple-selection question (MSQ). You record your answers on a computer-readable bubble sheet. Each question tells you how many answers to select (typically between two and six), and the correct answer set contains exactly that many items. Every question is weighted equally, contributing approximately 1.43% to your total score.
The scoring model is built around three principles:
- You earn marks for each correct answer you select (partial credit).
- You lose marks if you select more answers than requested (over-selection penalty).
- Incorrect selections within your allowed count are simply not credited, not penalised.
Understanding these three rules changes how you approach every KFP question. It turns answer selection from a guessing exercise into a calculated decision.
How partial credit works in the KFP
In a standard MCQ, you either get the question right or you do not. In the KFP, each correct answer within a question carries independent value. This means you are rewarded for every correct answer you identify, even if you miss some.
How partial credit works in practice
Consider a KFP question that asks you to select four management steps for an acute presentation. The question has four correct answers in its marking key.
- Select all 4 correctly = full marks for that question
- Select 3 correctly + 1 incorrect = credit for 3 correct answers
- Select 2 correctly + 2 incorrect = credit for 2 correct answers
- Select 0 correctly = zero marks
The incorrect answers you selected (within your allowed four) do not subtract from your score. They simply earn nothing.
This partial credit model has a direct practical implication: you should always select the full number of answers requested. If a question asks for four answers and you can only confidently identify two, you should still select two more using your best clinical judgement. Those two uncertain selections cannot hurt you if they are wrong, but they will earn marks if they happen to be right.
This is the opposite of the instinct many registrars have, which is to only select answers they are confident about. In the KFP, leaving a selection slot empty is the same as guaranteeing zero marks for that slot.
The over-selection penalty: where most marks are lost unnecessarily
The RACGP states this clearly in its exam conclusion:
Provide only the number of answers requested; providing additional answers will result in a penalty being applied to the overall score.
This is the single most important rule in the KFP, and the one most frequently violated. The over-selection penalty applies when you select more answers than the question asks for. Specifically, each extra bubble you shade beyond the requested number deducts 0.35% from your total exam score. If the question says "select four" and you select five, that single extra bubble costs you 0.35% of your overall result, regardless of whether the answer was clinically reasonable.
Why over-selection happens
The instinct to add an extra answer "just to be safe" is understandable. It comes from the logic of the AKT, where guessing carries no penalty. But in the KFP, the logic is reversed. Adding an answer beyond the requested count will always cost you, even if that answer is clinically correct. The penalty exists because the KFP is testing your ability to prioritise and select the most appropriateanswers, not to list everything that could theoretically be correct.
The practical rule is simple: read how many answers are requested, select exactly that number, then stop. Before you move to the next question, count your selections. If the number does not match what was asked, adjust.
What "key features" actually means in the KFP
The KFP is formally called the Key Feature Problem exam. The "key feature" concept is central to both the question design and the scoring. A key feature is the piece of information in the clinical scenario that changes the correct answer. Without it, the answer would be different.
The RACGP states this explicitly:
In the KFP exam, all information in the stem is given for a reason. Candidates should read the stems carefully, taking note of all key features, to ensure they receive maximal marks.
Key features are not just details. They are the pieces of information that constrain or redirect the correct clinical response. They include:
- Medications in the history that create contraindications
- Comorbidities that alter first-line management
- Age and demographics that change screening thresholds
- Location of the encounter (rural GP vs metropolitan hospital)
- Severity and acuity of the presentation
- Investigation results already provided in the stem
The scoring model rewards answers that account for key features. When candidates select a management step that ignores a contraindication clearly stated in the stem, that answer is not just "wrong" in academic terms. In the RACGP's framework, it demonstrates a failure to integrate clinical information, which is the core competency the KFP assesses.
Five worked examples from the 2026.1 exam report
The RACGP publishes annotated examples in every public exam report. These are not hypothetical. They describe real questions, real candidate errors, and the reasoning behind the scoring. Here are the five examples from the 2026.1 KFP report.
1 Acute STEMI with a contraindicated medication
The scenario
A 68-year-old man presents as a walk-in with acute ST-elevation MI. His medication list includes a phosphodiesterase 5 inhibitor (for erectile dysfunction). Candidates were asked to select four management answers.
What most candidates did well
The majority correctly administered aspirin, gave appropriate analgesia, and arranged urgent ambulance transfer to a tertiary hospital.
Where marks were lost
A large number of candidates also administered glyceryl trinitrate (GTN) spray. GTN is contraindicated with concurrent PDE5 inhibitor use due to the risk of profound hypotension. The RACGP described this as "a dangerous action" that "has the capacity to cause significant harm."
The scoring lesson
The PDE5 inhibitor was the key feature. Candidates who noticed it and excluded GTN from their four selections scored higher. Candidates who missed it selected GTN, which earned zero credit and displaced a potentially correct fourth answer.
2 IBD: selecting features that support vs exclude
The scenario
A 28-year-old woman presents with bloody diarrhoea. Key features indicated inflammatory bowel disease. Candidates were asked to select four history features that would support the provisional diagnosis.
Where marks were lost
Many candidates selected features associated with coeliac disease or irritable bowel syndrome instead of IBD. Others selected features that would exclude the diagnosis rather than support it.
The scoring lesson
This was a "two-step question" requiring candidates to first identify the likely diagnosis and then select supporting features. The question asked for features that support the diagnosis, not features that rule out alternatives. Reading the exact wording of the question is what separates full marks from partial marks.
3 Precocious puberty: appropriate initial investigations
The scenario
A 7-year-old girl with breast development and mood changes. A growth chart showed accelerating growth velocity. Candidates were asked to select five appropriate initial investigations.
Where marks were lost
Several candidates chose investigations appropriate for short stature or primary amenorrhoea rather than premature thelarche. The RACGP noted this suggested "a limited understanding of specific paediatric growth disorders."
The scoring lesson
The word "initial" was the key qualifier. High-performing candidates selected FSH, LH, estradiol, and a bone-age X-ray. They matched their investigations to the specific presentation rather than ordering a broad panel. Each correct investigation earned independent partial credit.
4 Off-label prescribing and PBS criteria
The scenario
A 45-year-old woman requests quetiapine on the PBS for insomnia and anxiety after being discharged from psychiatric services. Candidates were asked to select four management answers.
Where marks were lost
A common error was delegating the patient back to acute psychiatric services. This was an unnecessary referral for a stable patient who was appropriately managed in primary care. Many candidates also failed to recognise that the patient did not meet PBS criteria for quetiapine.
The scoring lesson
The KFP tests organisational and legal dimensions of general practice, not just clinical knowledge. Knowing PBS prescribing rules for common medications is examinable content. High-performing candidates recognised the off-label use, provided patient education, and incorporated CBT for insomnia (distinct from sleep hygiene).
5 Gout with CKD: when not to start treatment
The scenario
A 51-year-old man with an acutely inflamed first MTP joint, stage 3 CKD, and hydrochlorothiazide use. Candidates were asked to select four immediate next steps.
Where marks were lost
A common error was starting allopurinol before the diagnosis of gout had been confirmed. While current guidelines support starting allopurinol during an acute attack, the diagnosis must be established first. High-performing candidates also recognised the need to cease the thiazide diuretic.
The scoring lesson
The comorbidities (CKD, hydrochlorothiazide) were key features. The word "immediate" was the key qualifier. Candidates who matched their four selections to the specific clinical context earned partial credit for each correct answer. Those who selected plausible but context-inappropriate answers (like starting allopurinol without confirming the diagnosis) earned nothing for those slots.
Two-step questions: the format most candidates do not prepare for
The IBD example above illustrates a question type the RACGP calls a "two-step question." In this format, you cannot answer the question correctly without first arriving at the right diagnosis. If your provisional diagnosis is wrong, your supporting features or management selections will also be wrong.
Two-step questions test clinical reasoning depth, not just knowledge breadth. They require you to:
- Identify the most likely diagnosis from the clinical information provided.
- Select answers that are specific to that diagnosis, not to a related or similar condition.
The RACGP recommends using a diagnostic framework when approaching these questions. This means having a structured method for generating differentials and then testing them against the key features in the stem. The RACGP's own Exam Preparation Toolkit on gplearning includes guidance on diagnostic frameworks for this purpose.
In partial credit terms, two-step questions are high-stakes because a wrong first step (wrong diagnosis) usually means all subsequent selections are also wrong. But partial credit still applies: if you get the diagnosis right and select three of four correct supporting features, you earn credit for those three.
The scoring maths you should know
While the RACGP does not publish the exact mathematical formula used for KFP scoring, the public exam report provides enough data to understand the system in practical terms.
What earns marks
- Each correct selection earns independent credit
- Partially correct responses still earn marks
- Incorrect selections (within the allowed count) earn zero, not negative
- Filling every selection slot maximises your scoring potential
What costs marks
- Selecting more answers than requested costs 0.35% per extra bubble from your total score
- The penalty is deducted from your overall result, not just that question
- Leaving selection slots empty guarantees zero for those slots
- Answers the RACGP considers clinically dangerous (e.g. contraindicated medications) earn no credit
The 2026.1 KFP had a mean score of 71.57% against a pass mark of 63.29%. The gap between these two numbers is 8.28 percentage points. At a penalty of 0.35% per extra bubble, over-selecting on just 10 questions would cost 3.5% of your total score. For a candidate sitting near the pass mark, that alone could be the difference between passing and failing.
Practical takeaway
The candidates who fail the KFP are not, in most cases, failing because they lack clinical knowledge. They are failing because they lose marks through preventable scoring errors: over-selecting, not reading key features, answering a different question than the one asked, and leaving selection slots empty when they could have used clinical reasoning to make an educated guess.
What this means for your practice sessions
Understanding the scoring model should change how you practise KFP questions, not just how you sit the exam. Here are five specific adjustments to make.
Frequently asked questions
Is the KFP negatively marked?
Not in the traditional sense. Incorrect answers within your allowed selection count earn zero marks but do not subtract from your score. However, selecting more answers than requested triggers a penalty of 0.35% per extra bubble, deducted from your total exam score. The KFP has a targeted penalty for over-selection, not a blanket negative marking system.
Should I guess if I am unsure?
Yes. Always fill every selection slot. Within the allowed number of answers, incorrect selections do not penalise you. Leaving a slot empty guarantees zero marks for that slot. An educated guess has a positive expected value.
How much does each question contribute to the overall score?
Each of the 70 KFP questions is weighted equally, contributing approximately 1.43% to your total score. There are no "bonus" or "weighted" questions that count more than others. The pass mark is set using the modified Angoff standard-setting method.
Can I change my answers?
Yes, you can change answers before submitting the exam. If you realise you have over-selected, remove the answer you are least confident about. If you have under- selected, add your best guess. Counting your answers before moving on is the simplest way to catch these errors.
Practice with the real KFP format
Most question banks use single-best-answer MCQs for everything. The KFP is not an MCQ exam. Its partial credit model, over-selection penalty, and multi-selection format require a fundamentally different approach to practice.
FellowPath replicates the actual KFP scoring model: partial credit for partially correct responses, a real over-selection penalty, and analytics that tell you whether your errors are coming from knowledge gaps, misread questions, or scoring mechanics.
References
- RACGP. Exam Report 2026.1 KFP. East Melbourne, Vic: RACGP, 2026.
- RACGP. AKT and KFP Guide. racgp.org.au
- RACGP. Examination Guide. racgp.org.au
- RACGP. Exam Preparation Toolkit. gplearning (gpl.racgp.org.au)