MRCP PACES pass rate by year, 2018 to 2025 figures

Official PACES pass rates live in two places that disagree. One PDF per year in the annual report, and a per-diet web page with different numbers. This page pulls together the 2018 to 2025 annual figures in one place and explains why the sources diverge.

Boundary, before you ask. PACES examines physical signs in the respiratory, cardiovascular, neurology and abdominal encounters. Diagnosica, the AI patient simulator that publishes this blog, cannot rehearse any physical examination. Nothing here claims it changes anyone’s result. It’s reference material so you can see the figures without digging through seven PDFs.

Last reviewed: October 2026. The Federation updates the per-diet figures after each diet, so check its pages before relying on a number.

What is the MRCP PACES pass rate?

In 2025, 54.2% of UK Resident Doctor sittings passed PACES and 44.1% of other candidates' sittings passed. On first attempts, the rates were 56.6% and 43.0% respectively. These figures come from MRCP(UK) annual reporting of sittings and attempts, and they reflect self-declared training status.

MRCP(UK) labels UK Resident Doctors by self declaration. Everyone who did not declare training details is reported as other candidates in the same tables.

Across 2025 there were 7,485 sittings, 2,786 declared UK Resident Doctors and 4,699 other candidates. First attempt denominators were 1,591 UK Resident Doctor sittings and 2,481 sittings from other candidates. These are per sitting figures, not unique doctors.

Combined across all 7,485 sittings the pass rate was 47.9% (3584/7485), our arithmetic from the report's counts. All figures are taken from the 2025 annual report, produced by MRCP(UK) Assessment Data Analytics in February 2026.

PACES pass rates by year, 2018 to 2025

Here is every published MRCP PACES pass rate by year, split by group, with sittings.

• 2018. UK trainees: 59.0% first attempt, 55.9% all attempts. Other candidates: 38.5% first attempt, 37.5% all attempts. 6,530 sittings, from the 2018 report.
• 2019. UK trainees: 61.9% first attempt, 58.9% all attempts. Other candidates: 40.3% first attempt, 39.8% all attempts. 6,999 sittings, from the 2019 figures.
• 2021. UK trainees: 60.1% first attempt, 58.5% all attempts. Other candidates: 56.8% first attempt, 54.3% all attempts. 3,388 sittings, from the 2021 report.
• 2022. UK trainees: 58.1% first attempt, 57.1% all attempts. Other candidates: 46.5% first attempt, 45.1% all attempts. 6,376 sittings, from the 2022 figures.
• 2023. UK trainees: 57.2% first attempt, 55.0% all attempts. Other candidates: 45.0% first attempt, 44.8% all attempts. 6,904 sittings, from the 2023 report.
• 2024. UK Resident Doctors: 59.7% first attempt, 56.8% all attempts. Other candidates: 43.4% first attempt, 42.8% all attempts. 7,204 sittings, from the 2024 report.
• 2025. UK Resident Doctors: 56.6% first attempt, 54.2% all attempts. Other candidates: 43.0% first attempt, 44.1% all attempts. 7,485 sittings.

We couldn't find a 2020 annual report on the Federation's site, checked October 2026. The 2021 report counts 3,388 sittings, about half of 2019's 6,999.

Three definition changes matter when you compare years. The UK group label was "UK trainees" up to the 2023 report, then "UK Resident Doctors" from 2024. From 2022, UK trainees who self-declared as Other were excluded from the UK group. The PACES23 format arrived in the third diet of 2023.

Why do two official pages give two different numbers?

The annual report pools every sitting in a calendar year and splits UK Resident Doctors from everyone else, while the per-diet page reports one sitting at a time and uses different groups, including UK graduates. They measure different things, so they shouldn't be compared directly.

On the per-diet pass rates page, groups are "UK trainees 1st attempt", "UK trainees all attempts", "UK graduates 1st attempt", "UK graduates all attempts", and "All candidates". It currently shows the 2026/02, 2026/01 and 2025/03 diets only.

2026/02 diet: UK trainees 1st attempt 52.0%, UK trainees all attempts 53.2%. UK graduates 1st attempt 62.6%, UK graduates all attempts 64.8%. All candidates 47.9%.

2025/03 diet: UK trainees 1st attempt 51.4%, UK trainees all attempts 51.5%. UK graduates 1st attempt 65.8%, UK graduates all attempts 66.1%. All candidates 46.1%.

Older diets drop off that page, so if you are tracking a trend across years, you will miss past sittings unless you saved them.

The annual report aggregates all attempts in the calendar year and groups by UK Resident Doctors versus other candidates, who did not declare training details. Its counting rule is explicit, "Candidates will be counted multiple times if they have sat on more than one occasion." A doctor who failed at the second attempt and passed at the third appears once in each relevant column.

Both breakouts depend on self-reported status. The page caveat is clear, "UK trainee status is self-declared and therefore may not be fully representative of the UK trainee population."

First attempt vs resit: what the split actually says

Annual reports break the mrcp paces pass rate into three attempt bands, first attempt, attempts 2 to 6, and 7 or more. They count sittings not people, so the numbers here follow that convention.

For UK Resident Doctors the first attempt pass rate beats the 2 to 6 band every year shown, for 2025 it is 56.6% vs 51.1%, and for 2018 it is 59.0% vs 52.4%.

The same pattern holds in 2019 61.9% vs 54.7%, 2021 60.1% vs 56.0%, 2022 58.1% vs 55.7%, 2023 57.2% vs 52.0%, and 2024 59.7% vs 53.2%.

For other candidates in 2025 it goes the other way, attempts 2 to 6 at 45.5% beat first attempt at 43.0%. The 7 or more band is tiny, for 2025 UK it is 3 passes from 8 sittings, so do not read anything into it.

In the 2010 and 2011 diets, 347 candidates who sat PACES before passing Part 2 written passed at 57.6% vs 71.9% for those who had already passed it, from a 2013 Clinical Medicine study, over a decade old and the exam has changed format since.

If you are planning a resit, here is what comes after a fail.

The UK vs international candidate gap

On all attempts, the PACES pass rate gap between UK and other candidates was 18.4 percentage points in 2018 and 19.1 in 2019. It narrowed to 4.2 in 2021, then was 12.0 in 2022, 10.2 in 2023, 13.9 in 2024, and 10.1 in 2025.

Other candidates is the Federation’s category for sittings that did not declare UK training, so it is not the same thing as international graduates.

On first attempts in 2025, the pass rate was 56.6 percent for UK and 43.0 percent for other, a 13.6 point gap.

The Federation’s equality reporting says UK graduates and female candidates have higher PACES pass rates, and that white UK resident doctors have the highest PACES pass rate, with training, gender and ethnicity data self-declared, see the Federation's 2025 equality report.

The 2025 annual report also states that female candidates pass more often than male and White more often than BAME, for both UK resident doctors and other candidates, using all attempts.

The Federation points to a BMJ meta-analysis concluding non-UK trained doctors and students from minority ethnic groups tend to underperform academically compared with UK trained and white counterparts, but it is not clear why this is. That is where the published evidence stops. This page will not guess.

What the pass rate does not tell you about your own chances

A mrcp paces pass rate is a group average; it does not tell you if you personally clear the bar. You pass only if you meet every skill minimum and also reach the total score in the same sitting.

You must meet the seven skill floors and a total of at least 126 in one attempt: A Physical Examination 16, B Identifying Physical Signs 14, C Clinical Communication 11, D Differential Diagnosis 15, E Clinical Judgement 20, F Managing Patients' Concerns 10, G Maintaining Patient Welfare 28.

The MRCP(UK) Clinical Examining Board set the 126 pass mark on 28 May 2024, applying from the 2024/02 diet, after the PACES23 format began in 2023/03, see the published pass marks. The 2018, 2019, 2021 and 2022 annual reports state a minimum total score of 130.

So any cross year comparison of pass rates straddles both a format change and a different pass mark. Treat them as different eras.

In Spring 2025, the mean was 131 and the median 133, with most scores between 107 and 138. With the pass mark at 126 and the skills rule in force, a lot of attempts cluster near the line.

Practise daily Rehearse history-taking and communication with an AI patient, no physical examination content. Start a case free

Where candidates lose marks, and what you can rehearse away from the exam

The Federation’s own figures are blunt. Skill B, identifying physical signs, is the weakest, and Skill G, maintaining patient welfare, is the strongest, in the 2025 annual report and for all candidates in Spring 2025 where B was 66.2% and G 99.4% in the Spring 2025 diet report. Neurology sits furthest below its ceiling, the Communication encounters closest.

After B, the next soft spot for all candidates is Skill D, Differential Diagnosis at 68.6%, then Skill E, Clinical Judgement at 77.3% in the same Spring 2025 diet. If you’re hunting for the mrcp paces pass rate by skill, those two are the reasoning checks, not the signs hunt.

Be clear about B. It is a physical signs skill. It needs real patients, a tutor or a course. An app or a reading list will not fix it.

The Communication and Consultation encounters are different. Communications 1 had the highest proportional mean score at 86.4% for all candidates in Spring 2025, while Neurology was 74.3%. These are the domains you can rehearse without a clinic list, and there is a guide to PACES practice without a partner.

If you want a conversational partner to keep your history, explaining and concerns work ticking over, there are AI options for PACES prep. For how consultation stations are scored in the exam itself, see Station 5 consultation scoring.

PACES has eight patient encounters across five stations, with two Communication, two Consultation, plus Respiratory, Cardiovascular, Neurology and Abdominal.

Diagnosica for history and communication practice, and its limits

Diagnosica lets you take a history from an AI patient by voice or text, shown as a video avatar. You order investigations, see results and imaging, commit to a diagnosis and plan, then get a scorecard with competency scores and teaching points.

Every case is written and signed off by a doctor. Apps on iPhone and Android, plus the web app, with a free tier and no card.

"It doesn't give advice about real patients, and it's not a diagnostic system. Treat every output as educational."

There’s no physical examination practice of any kind, so it does nothing for skills A and B or the examination encounters. Skill B, identifying physical signs, is exactly what you can’t rehearse here.

It’s a general clinical reasoning simulator, not built around the PACES mark sheet, and it makes no claim about pass rates. The overlap is reasoning, including differential diagnosis, clinical judgement, history-taking and communication.

Educational use only. Not medical advice. AI-generated; verify clinically against primary sources.

Clinical review pending.

Quick warm up Try the two-minute quick case or a full case, no physical examination content. Try quick case