The MRCGP AKT changed in two ways recently, and most candidates noticed only one of them. The visible change is that from October 2025 the exam is 160 questions in 160 minutes, down from 200 in 190. The invisible change is that from February 2025 the RCGP uses Item Response Theory as its primary statistical analysis of item performance and standard maintenance. Neither changes what you need to know, but understanding both explains why raw percentages no longer compare across sittings and why chasing a score target is the wrong goal. Here is what happened and what it means.
Key takeaways
- From October 2025 the AKT is 160 questions in 160 minutes, with slightly more time per item.
- From February 2025 the RCGP uses IRT as its primary item analysis, with CTT retained in the background.
- The RCGP states the passing standard is unchanged, with no adverse demographic impact.
- IRT places item difficulty and candidate ability on a common scale, improving equating across sittings.
- The practical lesson is to chase mastery of weak domains, not a fixed raw-percentage target.
Two changes candidates noticed, and one they did not
Talk to a registrar who sat the AKT recently and they will mention the shorter paper. Far fewer will mention the statistical change underneath it, because the RCGP designed it to be invisible to candidates. Both are real, and they do different things: the format change alters your exam day, while the IRT change alters how the college analyses items and maintains the standard behind the scenes. Neither alters the content you revise, but the second one changes how you should interpret your practice scores.
The visible change: a shorter paper
From October 2025, the AKT moved from 200 questions in 190 minutes to 160 questions in 160 minutes, first sat as AKT 57 in October 2025. That gives you slightly more time per item, roughly one minute against the previous 57 seconds, which the RCGP frames as an inclusivity adjustment intended to reduce the speed-reading burden and to free up morning and afternoon sessions for candidates with reasonable adjustments. The college states the standard required to pass is unchanged, and that its submission to the GMC showed no adverse impact on any demographic group. Each question now carries proportionally more weight, but the content blueprint, 80 percent clinical, 10 percent evidence-based practice, and 10 percent organisation and management, is the same.
The invisible change: IRT
From February 2025, the RCGP began using Item Response Theory as the prime statistical analysis of item performance and the maintenance of the passing standard, while keeping Classical Test Theory running in the background to ensure a smooth, comparable transition. The college is explicit that this has no consequences for registrars: it does not change item content, curriculum coverage, the assessment of the passing standard, or the style of single-best-answer, multiple-best-answer, free-text, or extended-matching items. There is still one mark per correct answer and still no negative marking. It is a change to the machinery, not the exam you sit.
What IRT actually is
Classical analysis judges a question mainly by the raw proportion of candidates who got it right on a particular paper. Item Response Theory does something more powerful: it estimates each item's difficulty, and how well it discriminates between stronger and weaker candidates, and places that difficulty on the same scale as candidate ability. Because item and person sit on one common scale, IRT can compare performance across different papers and sittings in a way raw percentages cannot. It is the same family of models that underpins adaptive testing, though the AKT remains a linear exam, as we explain in how computer-adaptive testing works.
Why colleges make this change
The motivation is cleaner standard maintenance. With IRT, the college can equate difficulty across sittings more precisely, so the competence bar stays constant even when one paper is slightly harder than another, and even after shortening the paper from 200 items to 160. That is exactly the problem a professional exam must solve: different candidates sit different papers on different days, and the standard has to mean the same thing for all of them. IRT, combined with the Angoff standard-setting the RCGP continues to use, is how the college holds that line.
What it means for candidates
The practical consequence is about interpretation. Because the pass mark is set per sitting by standard-setting rather than fixed, and results are reported as scaled scores where zero equals the pass mark, comparing raw percentages across sittings is meaningless. In AKT 57, for example, the pass mark was 109 out of 160 and the overall pass rate was around 71 percent, but the raw pass percentage will differ next sitting because the paper's difficulty differs, while the underlying standard does not. So do not anchor on "I need 72 percent"; anchor on clearing the standard, which the college recalibrates each time.
Revision implications
If a fixed percentage is the wrong target, what is the right one? Mastery of your weak domains. Because the standard is held constant by equating and Angoff, the way to move from likely-fail to comfortable-pass is to close the specific gaps that cost you marks, not to grind your overall percentage up by a point. That means diagnosing which domains you are weak in, prioritising them, and retaining them, rather than doing undirected question volume. The iatroX engine uses the same family of models to target your weakest topics and schedule retention, and you can try it with free sample questions at iatroX. Chase the weak domain, not the round number.
Frequently asked questions
What changed in the AKT in October 2025? The exam moved from 200 questions in 190 minutes to 160 questions in 160 minutes, giving slightly more time per item. The content blueprint and marking are unchanged, and each question now carries proportionally more weight.
What is Item Response Theory in the AKT? It is the statistical model the RCGP has used as its primary item analysis since February 2025, placing item difficulty and candidate ability on a common scale so performance can be equated across different papers and sittings.
Did the AKT get easier or harder? Neither. The RCGP states the passing standard is unchanged and that there was no adverse demographic impact. The pass mark recalibrates each sitting to hold the standard constant despite paper-to-paper difficulty differences.
Why can't I compare AKT pass percentages between sittings? Because the pass mark is set per sitting by standard-setting, and results are scaled scores where zero is the pass mark. A raw percentage that passes one sitting may not another, since papers differ in difficulty while the standard does not.
How should I revise given these changes? Target mastery of your weak domains rather than a fixed percentage, since the standard is held constant by equating and Angoff. Diagnose your weak topics, prioritise them, and retain them with spaced practice.