PA Practice is a written-knowledge question bank for UK physician associates. For the Physician Associate Registration Assessment (PARA) it maps onto one component only — the knowledge-based assessment (KBA) — and cannot touch the 16-station OSCE that makes up the other half of the exam. As a first-pass breadth builder across the clinical matrix it is a reasonable option; used as your whole PARA plan it leaves the clinical component untested and its blueprint alignment unconfirmed.
What PA Practice offers for the PARA right now
The current-state box below records what the PA Practice website stated on the date checked. Every figure here is vendor-reported and should be re-checked on the product page before you buy, because at least one number is internally inconsistent.
| Item | What the vendor states (20 July 2026) |
|---|---|
| Live question count | "Over 3,000 questions" in one place; a footer states "over 2,500" — treat as approximate and verify the live count |
| Condition coverage | "Over 350 conditions from the physician associate clinical matrix", with disease summaries |
| Components supported | Written single-best-answer knowledge only; no OSCE or clinical-skills material |
| Adaptive/AI features | "Sophisticated performance analytics" to flag weak areas; no adaptive-difficulty or AI question-selection algorithm is claimed |
| Price | "Less than £5 per month"; exact subscription tiers and access period not clearly published — verify |
| Explicit PARA/GMC mapping | Not stated on the homepage checked; the bank is described around the PA clinical matrix rather than the RCP KBA blueprint |
The honesty flag to carry into every decision below: on the date checked, PA Practice presented itself as a physician associate revision bank built around the PA clinical matrix, and did not explicitly claim alignment to the GMC PARA content map or the Royal College of Physicians (RCP) KBA blueprint. That does not make the content wrong — the PA clinical matrix and the PARA blueprint overlap heavily — but it means you cannot take blueprint coverage on trust. You have to test it.
The PARA blueprint: what you are actually being tested on
The PARA is delivered by the RCP and, since the General Medical Council (GMC) began regulating physician associates in December 2024, it is a GMC-regulated assessment. It has two separate components. The KBA is a 200-question, single-best-answer paper delivered as four papers of 50 questions, one hour each. The OSCE contains 16 stations — 14 clinical scenarios plus two rest stations — with two minutes of reading and eight minutes inside each station. Candidates get up to four attempts at each component, and there is an 18-month window between passing the first component and the second.
Content is blueprinted against the 2023 Physician Associate Curriculum and its matrix of core clinical conditions. During the GMC transitional period (running to 13 December 2026 on the arrangements published at the time of writing) a hybrid blueprint spans both the 2012 Competence and Curriculum Framework and the 2023 curriculum; after that, the KBA and OSCE assess the 2023 curriculum only. The RCP does not publish a public per-domain numeric question weighting the way some royal-college membership exams do, so the honest unit of audit is condition coverage across the curriculum's clinical areas, not "so many cardiology questions".
Reading the headline number: coverage by clinical area
A three- or four-thousand-question total tells you almost nothing until you break it down. Do not audit PA Practice on its home-screen average; audit it against the curriculum's clinical areas. Build a simple matrix. Down the side, list the areas of clinical practice in the 2023 PA Curriculum — cardiovascular, respiratory, gastrointestinal, renal and urology, endocrine and metabolic, neurology, musculoskeletal, dermatology, ophthalmology, ENT, haematology, infectious disease, mental health, obstetrics and gynaecology, paediatrics, and the cross-cutting areas of pharmacology, safeguarding, ethics and prescribing safety. Across the top, put "questions attempted" and "first-pass accuracy".
Run 40 to 60 questions in each area and fill the grid in. The vendor's claim of "over 350 conditions from the physician associate clinical matrix" is a coverage promise you can falsify in an afternoon: if you can generate a full block in mental health, safeguarding and prescribing safety — the areas thin banks routinely under-serve — the promise largely holds for the KBA. If any area returns a shallow pool or recycles the same handful of stems, you have found a real gap, and no amount of overall percentage compensates for it. The completion-versus-coverage problem is the whole game here, and it is worth reading our blueprint-matrix method before you start so the grid is diagnostic rather than decorative.
Question style: recall, application and the clinical matrix
The KBA rewards application, not recall. A strong PARA item gives you a realistic primary-care or acute presentation, a short data set, and asks for the single best next step — the most likely diagnosis, the most appropriate first investigation, the safest immediate management. When you sample PA Practice, grade each item on four axes: stem length (one-line factual recall versus a genuine vignette), option plausibility (are the distractors all defensible, or is one obviously silly?), data and image use (does it interpret an ECG, a chest film, a set of bloods, or only describe them in text?), and management sequencing (does it ask what to do first, or merely what is true?).
Banks built primarily as condition summaries with attached questions tend to sit lower on this scale: correct, current, but more factual than the live KBA. That is not disqualifying for early revision — you need the facts before you can apply them — but it changes how you use the bank. If PA Practice items skew toward recall, it is a foundation-building and first-pass tool, and your final weeks must move to longer, application-heavy vignettes and mixed mocks that force next-step reasoning under time pressure.
Jurisdiction and recency: is the content current UK practice?
The PARA is a UK exam sat by physician associates working to UK guidance, so recency and jurisdiction matter as much as coverage. Take a stratified sample of 25 to 30 items across high-churn topics — hypertension thresholds, anticoagulation in atrial fibrillation, asthma and COPD management, sepsis recognition, type 2 diabetes escalation, and common antimicrobial choices — and check each answer against current NICE, CKS and, for medicines specifics, the SmPC on the electronic medicines compendium. Note the date you did this. A bank that still teaches a superseded target or an out-of-date first-line drug is a liability precisely in the areas the KBA loves to test.
Because PA Practice did not, on the date checked, publish a content-review date or a change log, treat currency as unverified until you have sampled it yourself. Anything you cannot reconcile with primary UK guidance should be parked and confirmed, not memorised.
The format gap: a question bank cannot sit your OSCE
This is the single most important line in the audit. The PARA OSCE is half the exam and PA Practice does not address it. Sixteen stations of history-taking, examination, communication, and management under a two-minutes-reading, eight-minutes-doing clock cannot be rehearsed by clicking single-best-answer options. A question bank builds the knowledge that underpins an OSCE station — you cannot manage a station on a topic you do not understand — but it does not build the performance: the structure, the timing, the safety-netting, the human interaction an examiner scores.
So state the boundary plainly to yourself. PA Practice, and any pure knowledge bank including iatroX, is a KBA tool and an OSCE-knowledge substrate, not an OSCE preparation product. Your OSCE plan needs live practice with simulated patients or peers, structured feedback against the station domains, and mock circuits. If a bank ever implies it "prepares you for PARA" without that caveat, discount the claim. For the reasoning behind why automated or self-scored feedback needs calibrating before you trust it, our note on calibrating AI-graded OSCE and SAQ feedback is the relevant read.
Duplication, near-duplicates and the recognition trap
Large banks grow by adding questions, and volume brings repetition. As you work through PA Practice, watch for three things: the same concept tested repeatedly with cosmetic changes, near-duplicate stems where only a number moves, and the slow conversion of "I know this" into "I remember this question". By your second pass, a familiar stem is answered from memory of the item, not from clinical reasoning — your accuracy climbs while your readiness does not. That is the recognition trap, and it is why bank percentage is not your exam score.
The defence is procedural. Keep at least one untouched, mixed, timed block in reserve for the final fortnight so you always have a clean readiness signal on unseen items. Reset your statistics between full passes so a 90% is 90% on fresh attempts, not on remembered ones. And when a topic starts feeling automatic, deliberately test it on a different bank to confirm the knowledge transfers rather than the wording.
Best-fit matrix: where PA Practice earns its place
| Use case | Fit | Why |
|---|---|---|
| Foundation building | Strong | Broad matrix coverage plus condition summaries suit early, systematic learning |
| First full pass | Strong | Large pool supports a complete run across clinical areas |
| Second/supplementary bank | Reasonable | Adds volume and fresh stems if your first bank thins in an area — mind duplication |
| Retake breadth | Reasonable | Useful to re-cover the matrix, but pair with a diagnosis of why you failed |
| Final KBA simulation | Partial | Only if items are genuinely vignette-style and timed; verify before relying on it |
| OSCE preparation | Not suitable | No OSCE material; this is a knowledge tool only |
A seven-day worked example
Here is one week for a physician associate eight weeks out, giving PA Practice a single defined job — systematic KBA breadth — and using iatroX for the separate job of unseen transfer measurement, without any claim to how either platform selects questions internally.
- Monday: PA Practice, one clinical area (say cardiovascular), 40 questions untimed, reading every explanation. Log the three weakest conditions.
- Tuesday: PA Practice, the next area (respiratory), same method. Add weak conditions to a running list.
- Wednesday: PA Practice, a deliberately awkward area you avoid — safeguarding, mental health or prescribing safety — 40 questions. These are the KBA's differentiators.
- Thursday: Consolidation. Re-test only Monday–Wednesday's flagged conditions, mixed, 30 questions.
- Friday: iatroX, a fresh timed mixed block on the week's areas — unseen items, no notes, KBA pace. This is your transfer test: can the knowledge survive a different question writer and a clock?
- Saturday: Reconcile. Any topic strong in PA Practice but weak on the iatroX block is recognition, not mastery — relearn it from guidance, not from the bank explanation.
- Sunday: Rest or a single 50-question timed paper to rehearse the one-hour, one-paper rhythm.
The point is the division of labour: one platform for structured coverage, a different, unseen bank for the readiness signal. Repeat weekly, rotating clinical areas, and keep one untouched mock for the final week.
Decision checklist: continue, supplement, switch or stop
- Continue if your matrix grid is filling evenly, first-pass accuracy on unseen mixed blocks is rising, and the content reconciles with current UK guidance.
- Supplement if one or two clinical areas return shallow pools or repetitive stems — add a second bank for those areas only, following the two-Q-bank rule so you do not duplicate practice or corrupt your percentage.
- Switch if a stratified sample throws up several out-of-date answers you cannot reconcile with NICE/CKS/SmPC, or if the items never rise above recall level as the KBA approaches.
- Stop using it for revision once accuracy plateaus on remembered questions rather than climbing on unseen ones — that is the signal to move to mixed mocks and, separately, to your OSCE circuit.
Every branch here is a measurable gap — coverage, currency, transfer, format — not novelty or sunk cost.
The bottom line
PA Practice is a credible written-knowledge bank for the PARA KBA: broad, condition-anchored, and cheap, with the honest caveats that its live question count is stated inconsistently, its explicit PARA/GMC-content-map alignment is unconfirmed, and its currency is undated. Verify those three things yourself, use it for breadth and first-pass coverage, measure readiness on unseen mixed blocks, and build an entirely separate plan for the OSCE it cannot touch. Do that and it earns a place in the stack; treat it as the whole plan and it will leave half the exam unrehearsed.
Frequently asked questions
Is PA Practice enough for the Physician Associate Registration Assessment on its own? No. It addresses the knowledge-based assessment only and does not provide OSCE or clinical-skills preparation, which is half of the PARA. Even for the KBA, its vendor-reported question count is stated inconsistently (around 3,000 in one place, 2,500 in another, verified 20 July 2026) and its explicit alignment to the GMC PARA content map is not published, so it should sit inside a stack that includes unseen mixed mocks and a dedicated OSCE plan rather than standing alone.
Which Physician Associate Registration Assessment component does PA Practice not reproduce well? The OSCE. The clinical component is 16 stations of history, examination, communication and management scored by examiners under a strict clock, and a single-best-answer question bank cannot rehearse that performance. PA Practice builds the underlying knowledge a station rests on, but it does not train the structure, timing or human interaction the OSCE assesses.
How many PA Practice questions should I complete per day for the Physician Associate Registration Assessment? There is no magic number, but a sustainable pattern for most candidates is 40 to 60 questions a day, always read to completion, with one topic covered in depth rather than a scattered high-volume dash. Quality of review beats raw count: a slower block where you understand every distractor moves you further than 150 rushed clicks, and it protects against turning revision into pattern recognition.
When should I stop using PA Practice and move to mixed mocks? Stop relying on it for new learning when your accuracy is climbing on remembered questions rather than on unseen ones — usually once you have completed a full pass and your second pass feels automatic. At that point your marginal gains come from timed, mixed, unseen mocks that rehearse the four-paper KBA format, and from your OSCE circuit, not from another sweep of familiar stems.
How should I combine PA Practice with iatroX without duplicating practice? Give each platform a distinct job. Use PA Practice for structured, topic-by-topic coverage of the clinical matrix, and use iatroX only for fresh, timed, mixed blocks that measure whether that knowledge transfers to unseen items at KBA pace. Never re-test the same items across both; the value of the second bank is that it is unseen. That separation follows the two-Q-bank rule and keeps each percentage meaningful.
Editorial notes and references
Written by Dr Kolawole Tytler, NHS GP and founder of iatroX. Last checked 20 July 2026. All PA Practice figures in this article are vendor-reported and were read from papractice.com on that date; at least one figure (the question count) was internally inconsistent, and no content-review date was published, so confirm the current numbers and currency on the product page before relying on them. Disclosure: iatroX operates a UK question bank that competes with PA Practice for the PARA knowledge layer; this audit confines iatroX's role to the unseen-measurement job that PA Practice does not claim, and takes no position on the OSCE preparation neither product provides. Corrections are welcome via the feedback route on iatrox.com.
References: RCP — PARA information for candidates (rcp.ac.uk); GMC — physician associate regulation and the 2023 Physician Associate Curriculum content map (gmc-uk.org); PA Practice product page (papractice.com), vendor-reported, 20 July 2026; iatroX — Your Q-Bank Percentage Is Not Your Exam Score; iatroX — question-bank completion is not coverage: building a blueprint-coverage matrix; iatroX PARA knowledge bank via quiz-landing and the comparison hub.