This is for physician associates using RCP Revise for the Physician Associate Registration Assessment (PARA) written exam who want it to keep measuring readiness rather than memory. It addresses the Knowledge-Based Assessment (KBA), not the OSCE. The principal limitation is counter-intuitive: RCP Revise is authored by the college that runs PARA, which makes its items unusually valuable as calibration — and that value is destroyed the moment you drill them to recognition. This protocol shows you how to spend that value once, well, rather than grinding it away.
What RCP Revise offers for PARA right now
Last checked 20 July 2026; figures are vendor-reported and change, so verify on the RCP store (store.rcplondon.ac.uk) before you buy.
| Item | What we could confirm (20 July 2026; verify) |
|---|---|
| Producer | The RCP — the body that delivers PARA |
| Format | Single-best-answer written questions mapped to the PARA KBA blueprint |
| PARA components covered | The written KBA only — not the OSCE |
| Question count | Not published in a confirmable form at check — verify on the RCP store |
| Price / access period | Not confirmed at check — verify on the RCP store |
| AI / adaptive features | None advertised — verify |
Because the RCP writes both the exam and these questions, RCP Revise items sit closer to the live paper's style than an adapted generic bank — the reason to treat them as calibration rather than as disposable drilling volume.
The PARA exam anchor
PARA is RCP-delivered and GMC-regulated (the GMC has regulated physician associates since December 2024). The written KBA is 200 single-best-answer questions across four papers of 50, one hour each. The June 2025 clinical-area weighting is led by Cardiovascular 8%, Acute and emergency care 7% and Gastrointestinal 7%, with the low-weighted domains (ophthalmology, palliative care, health promotion, healthcare service, research/statistics) at 2%; the cognitive weighting is Clinical management 40%, Diagnosis by investigation 20%, Diagnosis by interpretation 20% and Biomedical knowledge 20%. The OSCE is a separate 16-station exam. The genuinely official calibration material — blueprints, released sample questions, regulations and the annual report — is finite; RCP Revise is the RCP's larger, refreshable revision bank. Verify current details on rcp.ac.uk.
Why RCP-authored items are calibration gold — and why grinding them wastes them
Calibration is the act of using material of known fidelity, unseen, to read your true position. RCP Revise items are high-fidelity by provenance. But fidelity only helps while the item is unseen: once you have seen a question two or three times, your response reflects memory of that question, not your ability to reason through a new one. Drilling RCP Revise to a high percentage therefore feels like progress while actively converting your best calibration instrument into a recognition test. The whole protocol below exists to prevent that.
Inventory and label every item
Before you study, inventory your official and RCP-authored material and label each item's state. The label determines how you are allowed to use it.
| Label | Meaning | Permitted use |
|---|---|---|
| Unseen | Never attempted | Reserve for calibration — sit once, timed |
| Attempted once | Sat under exam conditions, reviewed | Calibration spent; move learning to transfer questions |
| Contaminated by review | Re-read repeatedly; answer now recognised | No calibration value; use only to teach the principle, then test elsewhere |
The discipline is to keep the "unseen" pool genuinely unseen until you deliberately spend it — not to nibble at it during casual study.
Choose the calibration date
Pick the date you will spend a calibration set deliberately: late enough that it is meaningful — after your first blueprint pass — but early enough that you can still act on what it reveals. Sitting your calibration too early wastes it on an unready version of you; sitting it the night before leaves no time to correct the weaknesses it exposes. For most candidates a calibration point around the two-thirds mark of their preparation, with a second near the end using genuinely fresh material, works well.
Reproduce exam conditions exactly
When you spend a calibration set, reproduce the exam precisely: the timing (roughly one minute per question, in a 50-item paper structure), the allowed references (none beyond what the exam permits), the breaks, and the response format. A calibration sat with notes open, untimed, or in fragments tells you nothing about exam-day performance. The point is to measure the version of you that will sit in the actual centre, not a relaxed study version.
Code every error by domain, cognitive process and format
After the sitting, code every error along three axes — not merely by subject. Subject alone ("I got a cardiology one wrong") is too coarse to act on.
| Axis | Codes |
|---|---|
| Domain | The blueprint clinical area (cardiovascular, acute care, GI, …) |
| Cognitive process | Knowledge, interpretation, management, premature closure |
| Format | Long-stem reading, data interpretation, calculation, timing |
Coding this way turns a raw score into a set of specific, fixable instructions — for instance, "management errors in acute care under time pressure", which points to a very different fix from "knowledge gaps in dermatology".
Map errors to fresh third-party practice
Now the crucial move: map each coded error to fresh practice, and keep the official and RCP-authored questions out of your daily repetition. If your calibration exposed acute-care management errors, drill acute-care management on a different bank — iatroX or another third-party source — with new items testing the same principles. The official material stays clean for the next calibration; the daily grind happens on renewable questions. This is exactly the division of labour behind the two-Q-bank rule: one bank for measurement, another for volume.
Repeat only with genuinely unseen material
Re-calibrate only with material you have not seen — a fresh released sample, or a portion of RCP Revise you ring-fenced and never touched. If no genuinely unseen official-adjacent material remains, do not re-sit old items and call it calibration; use transfer questions on a different bank instead and accept that your remaining calibration reads come from unseen third-party mocks. Honesty about what is unseen is what keeps the signal meaningful.
Seven-day worked example
One week preserving RCP Revise as calibration while iatroX carries the daily unseen drilling. No proprietary-algorithm claims — you run the protocol.
| Day | RCP Revise job | iatroX job (unseen measurement/drill) |
|---|---|---|
| Mon | Spend one calibration set (unseen, timed, exam conditions) | — |
| Tue | Three-axis error-coding of Monday's misses | Fresh items on the top two error clusters |
| Wed | — (kept clean) | Transfer drilling on weak management areas |
| Thu | — (kept clean) | 20 unseen mixed items, timed |
| Fri | Review error log; plan the fixes | Targeted items on the dominant format error |
| Sat | — (kept clean) | Mock-style mixed block, timed |
| Sun | Decide the next calibration date | 25 unseen mixed items as a readiness read |
Note that RCP Revise is touched on only two days — once to calibrate, once to code — and never drilled to familiarity. iatroX carries the repetition so the calibration instrument stays sharp.
Three mistakes this protocol is designed to stop
First, drilling RCP Revise to a high percentage and mistaking recognition for readiness. Second, spending your calibration too early or too late to act on it. Third, re-sitting seen items and calling it calibration, which measures memory and quietly inflates your confidence. A familiar-bank percentage is not a pass prediction, and a finished bank can still leave blueprint gaps.
Decision checklist: continue, supplement, switch or stop
| Signal | Action |
|---|---|
| Unseen calibration material remains; error codes are shrinking | Continue |
| Errors cluster in specific domains or formats | Supplement with fresh third-party drilling there |
| RCP Revise items are becoming familiar | Switch daily practice fully to a fresh bank; ring-fence any unseen remainder |
| Coverage floors met, pace stable, latest calibration holds | Stop adding material; consolidate and rest |
Every decision follows a measured gap — an error cluster, a coverage floor, a pacing figure — not novelty or the sunk cost of the subscription.
Bottom line
RCP Revise is a strong PARA option precisely because the exam's own college writes it, which is also why it must be protected. Label every item, spend your calibration set once under exam conditions at a date you can act on, code the errors three ways, and route all daily repetition through fresh questions so the RCP-authored items keep telling you the truth about your readiness.
Frequently asked questions
Is RCP Revise enough for PARA on its own? For the written KBA it is a strong, blueprint-faithful core because the RCP authors it, but on its own it is best treated as your calibration and high-fidelity practice reference rather than your entire study — you still need a renewable source of fresh questions for daily drilling and separate OSCE preparation. Whether it holds enough unseen volume depends on a question count we could not confirm; verify it on the RCP store so you know how long your unseen pool will last.
Which PARA component does RCP Revise not reproduce well? The OSCE. RCP Revise is a written single-best-answer resource for the knowledge component and does not prepare the 16-station clinical exam, its communication and procedural domains, or examiner-observed timing. Use OSCE-specific preparation for that component and treat RCP Revise — like iatroX — as covering only the written knowledge layer.
How many RCP Revise questions should I complete per day for PARA? Under this protocol, deliberately few — RCP Revise is your calibration instrument, not your daily grind, so you might touch it only on calibration and review days rather than every day. The bulk of your daily questions (30–60, with review) should come from a renewable bank such as iatroX, keeping the RCP-authored items unseen until you spend them on purpose. Protect the pool rather than working through it steadily.
When should I stop using RCP Revise and move to mixed mocks? You should already be doing mixed mocks on other banks throughout; RCP Revise's role is periodic calibration, so you "stop" using it when you have spent your genuinely unseen calibration sets and have no fresh official-adjacent material left. At that point, rely on unseen third-party mocks for your remaining readiness reads and keep any released official sample for a final calibration close to the exam.
How should I combine RCP Revise with iatroX without duplicating practice? Assign them opposite jobs: RCP Revise is your protected, high-fidelity calibration reference; iatroX is your renewable daily-drilling and unseen-measurement bank. Never move RCP Revise items into daily repetition or re-answer them across tools to pad a percentage — that destroys their calibration value. When calibration exposes a weakness, drill it with new iatroX items on the same principle, exactly as the two-Q-bank rule prescribes.
Editorial notes and references
Written by Dr Kolawole Tytler, NHS GP and founder of iatroX. Last checked 20 July 2026. RCP Revise's question count, price and access period were not confirmable at check; verify on the RCP store (store.rcplondon.ac.uk) before purchase, and note that RCP Revise is a revision product, not the exam papers. Disclosure: iatroX operates a competing UK question bank and covers the PARA written/SBA layer only — not the OSCE and not the RCP's own product; this article confines iatroX's role to renewable daily drilling and unseen measurement, the job a protected calibration bank should not be used for. Corrections via the feedback route on iatrox.com.
References: RCP — PARA information for candidates and KBA blueprint (June 2025), rcp.ac.uk; GMC — PA registration assessment content map, gmc-uk.org; RCP Revise Question Bank for Physician Associates, store.rcplondon.ac.uk. Internal: Your Q-Bank Percentage Is Not Your Exam Score; the two-Q-bank rule; the iatroX comparison hub.