Pass rates are the most quoted and most misread numbers in medical assessment, pressed into service as difficulty rankings, quality verdicts and personal probability estimates, jobs they cannot do. This page is the methodology hub for every pass-rate analysis on this site: the concepts that make official statistics readable, the inferences they genuinely support, and the popular conclusions they do not. It exists so that our data investigations, PLAB, ORE, NDEB and the rest, can analyse numbers without re-teaching the foundations each time, and so candidates can use official figures as information rather than folklore.
Pass rate versus pass mark, raw versus equated
Four terms do most of the work. The pass mark is the standard a candidate must reach; the pass rate is the proportion of a cohort that reached it; conflating them produces most amateur analysis. Raw marks are counted answers; scaled or equated scores place performance on a common measurement scale so that standards hold constant while paper difficulty varies, which is why modern examinations report figures like MRCP Part 1's 450 that deliberately resist percentage translation: /blog/mrcp-part-1-pass-mark-450-explained. Standard setting is how the pass mark is determined, criterion-referenced methods anchor it to defined competence rather than to a quota, which is why a whole cohort can in principle pass, and why a falling pass rate is not evidence that examiners tightened a quota. And equating is why "this diet was easier" folklore is largely wasted energy: the mechanism exists precisely to neutralise it, as the AMC's 2026 adjustment illustrates, a slight increase in the MCQ passing standard while results continue to be reported on the familiar 0 to 500 scale with the pass mark described as 250, a change that rewards careful explanation rather than headline arithmetic.
The cohort problem: composition explains more than difficulty
The single most under-appreciated driver of pass-rate movement is who sat, not what was asked. Cohorts shift in first-attempt versus repeat mix, and repeat candidates are a selected population whose pass rates run structurally lower; they shift in training background, preparation time and route into the examination; and they shift in size, with policy changes, visa flows and scheduling backlogs moving thousands of candidates between years. A pass rate can therefore fall while the examination and the standard stay identical, purely through composition, and regulators frequently say so in their own reporting; the honest analyst's first question about any change is always "did the cohort change?", and only after exhausting that does difficulty deserve the stage.
Volatility, denominators and the small-cohort trap
Percentages detach readers from sample sizes, and small cohorts make wild percentages routine: a specialty examination sat by forty candidates moves ten points on the performance of four people, which is noise wearing a trend's clothing. Reading rules that prevent embarrassment: always recover the denominator; treat single-year movements in small cohorts as weather, not climate, and prefer multi-year weighted views; expect volatility to scale inversely with cohort size when comparing specialties; and beware comparisons across years where the examination's format, blueprint or candidate categorisation changed, because a redefinition can manufacture a trend out of continuity. These are exactly the disciplines our specialty-level analyses apply, and the reason they publish denominators and uncertainty warnings alongside every rate.
What you may and may not conclude
May conclude, with care: how a defined cohort performed against a defined standard at a defined time; multi-year direction of travel in large, stable cohorts; and where your own attempt sits relative to a base rate, as context, not destiny. May not conclude: that a low pass rate proves an examination is "hardest", since candidate selection and standard-setting differ across examinations; that a high rate proves easiness, for the mirror-image reasons; that a falling rate proves the examination got harder, before cohort effects are excluded; that any preparation resource caused an aggregate movement, a claim aggregate data cannot support in either direction; or your personal probability, because you are not a random draw from the cohort, and preparation, attempt number and background move individuals far more than base rates do. The practical use of official statistics is planning, respect the examination, budget genuine preparation time, take base rates as seriousness signals, and then let your own unseen-question performance, not the cohort's history, tell you when you are ready.
Frequently asked questions
Why do regulators publish pass rates at all if they are so misread?
Transparency and accountability: the figures let standards be scrutinised and trends be questioned, which is healthy; the misreading is downstream, and pages like this one are the repair.
Is there any examination where the pass rate is a quota?
Norm-referenced elements exist in some selection processes, but the major licensing and membership examinations covered on this site set criterion-referenced standards, which is exactly why pass rates can move while standards do not.
What single habit most improves pass-rate literacy?
Recovering the denominator and the cohort description before reacting to the percentage; thirty seconds of context defuses most bad takes before they form.