If you are asking which MRCGP AKT resource is best, the honest answer is that there is no single best bank; there is a best resource for your profile, your weeks to exam and your budget. This decision tree is for ST3 GP trainees choosing what to buy and what to do next. It replaces "which is the most popular Q-bank" with a set of measurable branch criteria, a platform-to-job matrix and three worked plans. Its one limitation: it depends on you taking an honest, timed unseen baseline first.
How to use this decision tree
Do not start by buying anything. Start by producing a domain profile: sit a small, timed, unseen sample stratified across the AKT's three areas, then read your first-attempt score by domain rather than your headline percentage. Every branch below keys off that profile plus two numbers you already know: weeks to your chosen sitting and the budget you can actually spend. Novelty and sunk cost are explicitly excluded as decision criteria. If you have already paid for a resource that no longer matches your profile, that is a reason to change what you do with it, not a reason to keep feeding it.
The exam you are buying for
From October 2025 the MRCGP Applied Knowledge Test is 160 single-best-answer questions in 2 hours 40 minutes (reduced from 200 in 3 hours 10 minutes), delivered at Pearson VUE test centres across four sittings a year. The blueprint weighting is approximately 80% clinical medicine, 10% evidence-based practice (statistics, critical appraisal and research understanding) and 10% health informatics and organisational general practice (UK administration, regulation and systems). Every item is set in a UK general-practice context, and the pace is roughly one minute per question. Two features of that blueprint drive most resource choices: the two 10% minority domains are where broad clinical banks are thinnest, and the UK-GP framing means content currency (NICE, CKS and current SmPC/eMC medicines information) matters as much as raw volume.
The minimum stack
Most candidates over-buy. The minimum effective stack for the AKT is four slots, and two of them are often empty:
- One primary question bank for volume and coverage. This is your workhorse; you should not run two full banks in parallel.
- Official calibration material — the RCGP's AKT example questions and content guide. This is finite, so you calibrate format and pacing against it rather than train on it.
- One teaching or reference source, only where a domain is genuinely weak. A video course, a statistics primer or a clinical-update product belongs here, not by default.
- One modality or measurement tool where relevant — for most candidates this is a source of fresh, unseen items to test transfer and to run the two-Q-bank rule without duplicating questions.
If your baseline profile is even and your knowledge is sound, slots three and four stay small. Adding resources you do not need dilutes retrieval time; it does not add coverage.
Segment yourself first
| Profile | Defining signal | Principal risk |
|---|---|---|
| First attempt | No prior AKT data; uneven baseline | Buying volume before knowing where the gaps are |
| Retake | A recorded fail with a domain breakdown | Re-running the same bank and re-memorising seen items |
| Busy trainee (revising around clinical work) | Under 6–8 study hours a week | Passive consumption crowding out timed retrieval |
| Weak foundations | Multiple domains below mid-band | Jumping to mocks before the teaching is done |
| Strong knowledge, poor pacing | High untimed accuracy, unfinished mocks | Endless topic-filtered practice, never timed |
| Strong clinical recall, weak applied | Clinical ≥ good, EBP/organisational low | Ignoring the two 10% domains that are easiest to lift |
Budget bands
Verify every figure on the vendor's page on the day you buy; the numbers below are vendor-reported and dated 19 July 2026.
- Free / low-cost: the RCGP example questions, a free UK-core bank (iatroX offers one), and NHS and deanery AKT resources. Enough to calibrate and to expose gaps; rarely enough volume alone.
- One premium resource (roughly £15–£130): a single established AKT Q-bank subscription, or one structured course such as the FourteenFish AKT Package (vendor-reported £95 including VAT for 12 months). Choose the type that fixes your weakest slot, not the best-known brand.
- Comprehensive stack (£150+): premium bank plus a course plus live teaching or a mock series. Justified for retakers, weak-foundation candidates and anyone who has failed once, not for a strong first-timer.
Time bands: decide what to omit
The mistake under time pressure is to add; the discipline is to omit.
- More than 12 weeks: you can afford a teaching phase. Sequence course or reading first, then bank volume, then timed mocks. Omit nothing, but front-load the weak domain.
- Four to 12 weeks: run bank volume and spaced error review in parallel from day one; use the course only for your single weakest domain. Omit any second full bank.
- Under four weeks: triage. Timed mixed mocks plus highest-yield clinical topics plus the two minority domains. Omit new courses, omit starting a fresh full bank, omit note-making you will not revisit.
The platform-to-job matrix
Map each resource to the one job it does best, then buy only the jobs your profile needs. This article is the hub; the detailed evidence for each platform lives in its own audit, so the summary stays deliberately short here.
| Resource | Best single job | Note (verify on the product page) |
|---|---|---|
| RCGP AKT example questions and content guide | Official format and pacing calibration | Finite; the gold standard for phrasing, not for volume |
| Passmedicine (AKT) | High-volume core bank and value | Broad coverage, concise explanations; confirm current count/price |
| Pastest (AKT) | Full timed mocks and explanation depth | Exam realism; confirm current count/price |
| FourteenFish AKT Package | Structured teaching and clinical updates | Video course plus one mock; not a high-volume bank |
| Arora, Emedica and similar courses | Live teaching and accountability | Timetabled structure; confirm current fees and format |
| iatroX MRCGP AKT bank | Unseen measurement and AI verification | Free UK-core bank and Socratic Tutor, NICE/CKS-grounded |
Note that "simulation" is not an AKT job at all — consultation realism belongs to the separate MRCGP SCA. For the AKT, the nearest equivalent is a full timed mock under exam conditions.
Cannibalisation guardrail
This page deliberately does not re-run each platform's full audit. If you want the grounding, feedback and coverage evidence for a specific tool, follow the child audit for that product and the completion-is-not-coverage blueprint method rather than trusting the one-line summary above. The purpose here is the decision, not the dossier.
The decision table
| If your baseline shows… | Weeks to exam | Recommended branch |
|---|---|---|
| Even profile, every domain at or above mid-band | Any | One high-volume bank plus RCGP calibration; add unseen measurement only to confirm transfer |
| One domain far below (e.g. EBP/statistics under 50%) | 8 or more | Add a targeted teaching source for that domain; retest on unseen items, not the same block |
| Multiple domains low (broad weakness) | 12 or more | Structured course first, then bank volume, then timed mocks |
| Good knowledge but unfinished, timed mocks | Any | Timed mixed mocks only; stop topic-filtered practice |
| Recorded fail with a weak-domain breakdown | 8 or more | Switch primary bank to avoid seen items; official calibration; unseen measurement to confirm change |
| Incomplete preparation, short runway | Under 4 | Triage: timed mixed mocks plus the two minority domains; buy no new course |
Three worked candidates
Priya — first attempt, 10 weeks, moderate budget. Baseline (timed, unseen, 60 items): clinical 72%, EBP/statistics 48%, organisational 55%. Even clinical knowledge but two soft minority domains. Branch: one high-volume bank plus a short statistics resource; no full course. Weekly shape — four clinical bank blocks (40 items, timed), one statistics/EBP session, one organisational block, spaced review of every miss by error code, and a fortnightly 60-item unseen block to confirm the minority domains are moving. RCGP example questions in weeks 8 and 9 for pacing. Exit criteria: EBP and organisational both above 65% on unseen items, clinical stable above 75%, mocks finished inside time.
Tom — retake, 8 weeks, previous fail. Feedback showed organisational and EBP below the passing standard; clinical adequate. Branch: switch primary bank so he is answering unseen items rather than re-recognising a bank he has already exhausted, add FourteenFish clinical updates for currency, and use unseen measurement weekly to prove the change is real rather than a familiarity effect. Weekly shape — two organisational and two EBP blocks, two mixed clinical maintenance blocks, one full timed mock, RCGP calibration in the final fortnight. Exit criteria: both failed domains above 65% on unseen items across two separate weeks, not one lucky block.
Sara — busy trainee, under 4 weeks, strong knowledge, poor pacing. Untimed accuracy is high; she runs out of time and leaves items blank. Branch: no new resource. Daily timed mixed mocks at exam pace, a hard rule to flag and move on rather than dwell, and unseen timed blocks to check the pacing fix transfers to fresh material. Exit criteria: two consecutive full mocks completed inside 2 hours 40 minutes with accuracy maintained. If pacing is stable a week out, she stops adding questions and rests.
The evidence hierarchy behind these choices
Rank your sources deliberately. Use the official RCGP material first for format, pacing and blueprint weighting. Use primary UK guidance — NICE, CKS and current SmPC/eMC medicines information — for clinical content and currency. Use vendor pages for product facts such as question counts and prices, treating them as vendor-reported and dating them. Use independent testimony and reviews last, and only for user experience, never for clinical truth. When two sources disagree, the higher tier wins.
Bottom line
There is no universal best AKT resource, only a best next action for your profile. Take a timed unseen baseline, read it by domain, then follow one branch: even and strong means one bank plus calibration; a single weak domain means a targeted teaching add-on; broad weakness means teaching before volume; a pacing problem means timed mocks only; a retake means a fresh bank plus proof of change. Buy the job you are missing, not the brand you have heard of, and let unseen measurement — not bank completion — tell you when to stop.
Frequently asked questions
How do I know whether I have covered the full MRCGP AKT blueprint? Completion of a bank is not coverage of the blueprint. Build a coverage matrix that lists the AKT's three areas — clinical medicine, evidence-based practice and organisational general practice — and record your unseen first-attempt performance in each, not how many questions you have ticked off. You have covered the blueprint when every cell has an adequate sample of timed unseen items behind it and none is sitting far below the others, regardless of your overall percentage.
Can one question bank be enough for MRCGP AKT? For a candidate with sound foundations and an even domain profile, one high-volume bank plus the official RCGP calibration material can be enough, and adding a second bank early simply duplicates content and inflates your familiar-item percentage. One bank stops being enough when you have exhausted its unseen items, when a minority domain stays weak, or when you are a retaker who has already memorised much of it — at which point a second, unseen source becomes a measurement tool rather than more of the same.
What should I measure instead of my overall Q-bank percentage for MRCGP AKT? Measure your first-attempt accuracy on unseen, timed items broken down by the three blueprint areas, your completion rate inside the time limit, and your retention of previously missed items when they reappear weeks later. A rising overall percentage driven by re-seeing familiar questions is the single most misleading number in exam preparation; your unseen, timed, domain-level score is the one that tracks the real exam.
When should I stop doing new MRCGP AKT questions? Stop adding new questions when every blueprint domain is at or above your target on unseen timed blocks across two separate weeks, your mocks finish inside time, and your error rate is dominated by careless slips rather than knowledge gaps. Beyond that point, more new questions add fatigue rather than coverage; the higher-value activity is spaced review of your existing misses and rest before the sitting.
Which MRCGP AKT resource should I use for my weakest component? Match the resource to the component: for weak clinical breadth, a high-volume bank; for weak evidence-based practice, a focused statistics and critical-appraisal source plus targeted questions; for weak organisational knowledge, current RCGP and NHS administrative material rather than a clinical bank, which typically under-covers it. In every case, confirm the fix on unseen items in that specific domain rather than assuming the resource worked because you enjoyed it.
Editorial notes and references
Written by Dr Kolawole Tytler, NHS GP and founder of iatroX. Last checked 19 July 2026; all platform figures, prices and features are vendor-reported as at that date and should be re-verified on each product's page before you buy, because counts and prices change without notice. Disclosure: iatroX operates a competing MRCGP AKT question bank; its role in this article is confined to unseen measurement and format calibration, the jobs a single existing bank or a video course does not claim to do. Corrections are welcome via the feedback route on iatrox.com.
References: RCGP — Applied Knowledge Test (introduction, preparing and example questions), rcgp.org.uk; vendor product pages for Passmedicine, Pastest and the FourteenFish AKT Package; iatroX MRCGP AKT bank, https://www.iatrox.com/mrcgp-akt; "Your Q-Bank Percentage Is Not Your Exam Score", https://www.iatrox.com/blog/qbank-percentage-not-your-exam-score; the completion-is-not-coverage blueprint method, https://www.iatrox.com/blog/question-bank-completion-is-not-coverage-how-to-build-a-blueprint-coverage-matrix-for-any-medical-exam; and the iatroX comparison hub, https://www.iatrox.com/compare.