This article is adapted from a post in Dr Koshy Jacob's "Building Revise Radiology in Public" series on LinkedIn.

Almost every candidate who fails the FRCR 2B arrives at the same explanation within a fortnight, which is that they did not know enough radiology. Sometimes that is true. More often it is one factor among several, and the others are more specific, more addressable, and considerably less damaging to a person's sense of their own ability.

The useful thing about this question is that we are not required to speculate. The Royal College publishes an examiners' report after every sitting, in which the Examination Board sets out, often in remarkable detail, exactly where candidates lost marks. These reports are freely available, they are written expressly to help candidates and their trainers, and my strong impression is that most candidates have never read one.

One disclosure before I go further. Some years after failing 2B myself, I ended up as one of the co-authors of a survey published in Clinical Radiology which asked 258 recent candidates how they had actually experienced the examination. That paper supplies several figures below, and a reader is entitled to know when the person citing a study had a hand in writing it.

The numbers, and how much they move

The College's own reports give sitting-by-sitting pass rates, and the range is wider than most candidates realise. Across six consecutive sittings the overall pass rate ran at 82.9%, then 60%, then 48.5%, then 40%, then 68%, then 67%. In one sitting the UK pass rate was 42.9%. In another it was 84.5%.

Before anyone concludes the examination is capricious, look at the composition. The strongest sittings had a high proportion of UK trainees, and the weakest had several times more global candidates than UK ones. The variance tracks who was sitting rather than how hard the paper was. But it does mean something important for an individual: the raw probability attached to your sitting was not the same as the one attached to somebody else's, and a bare pass rate quoted at you by a colleague may describe a quite different cohort from your own.

The examination has changed, and a lot of the advice has not

This is the most practically urgent thing in this article. Rapid reporting no longer exists. It has been replaced by short case reporting, and the two are different examinations testing different things.

Rapid reporting asked for 30 radiographs in 35 minutes, a mixture of normal and abnormal, with roughly 90% accuracy required to pass. It was explicitly a test of speed and of the nerve to call a film normal. Short case reporting asks for 25 cases, marked out of five each for a total of 125, with no half marks, double marked by two examiners independently and the scores averaged. Every case carries specific marking guidance covering the key findings, the diagnosis and the recommended onward management. Roughly half to sixty per cent are chest, forty to fifty per cent musculoskeletal, with about one abdominal case. Around a quarter are paediatric.

And, crucially, every case is abnormal.

Sit with that for a moment, because it inverts the old examination completely. Under rapid reporting, the single most reliable finding in the examiners' reports was that unsuccessful candidates overcalled: they saw abnormality in normal films and it cost them. The best-known piece of candidate folklore, passed down through blogs and study groups, was that you should overcall rather than miss, since a reasonable false positive still earned you half a mark. That advice was arguably wrong then and it is meaningless now. There are no normals to overcall, no half marks to salvage, and the task is not deciding whether something is there but finding it, describing it precisely, and saying what should happen next.

If your preparation material or your mentor's advice predates this change, some of it is now actively unhelpful.

Why short cases are hard, for entirely new reasons

The old component punished poor discrimination and slow reading. The new one punishes incomplete reporting, and there are far more places to leak marks.

Five marks per case means each short case is closer to a miniature long case than to a snap call. You can identify the abnormality correctly and still fall well short of five, because the mark scheme also wants accurate characterisation and the appropriate next step. Management has therefore migrated into all three components of the examination, and the examiners have said repeatedly, across many years of reports, that management must go beyond a referral to the multidisciplinary meeting. What they want is the advice you would give that meeting.

The images are static. You cannot scroll, and you cannot window or level your way out of uncertainty, which removes a crutch that daily practice quietly provides. The format also explicitly permits more complex imaging than the old rapid reporting did, so the ceiling has risen.

Then there is the case mix, which deserves more attention than it gets. Around a quarter of the component is paediatric, and paediatric plain film interpretation is the single most frequently repeated weakness in the entire archive of examiners' reports. Sitting after sitting the Board notes that candidates' knowledge of normal paediatric appearances is poor. The new format takes the most commonly documented deficiency in the candidate population and gives it a quarter of a component. If you are looking for the highest-yield thing to fix, it is probably that.

Precision remains everything. The reports give worked examples that are worth studying properly. Reporting a fracture without saying where earns nothing, because the examiner cannot be sure you found the right abnormality. Naming the bone earns something. Naming the bone, the site and the fracture type earns full credit. Get the side wrong on a paired structure and you lose the marks entirely, where simply omitting the side would have cost you less. Where a fracture is pathological, the underlying lesion has to be described as well as the fracture. Where a structure normally breaks in two places, or forms a ring, both injuries are expected.

There is also a cautionary tale in the reports about interface design. When the written platform gained an unlimited text box, candidates began writing longer answers, those answers became less precise, more half marks were awarded, and the Board recorded that a greater number of candidates consequently failed that component. It had to adjust the mark conversion. Nobody's radiology got worse. The box got bigger.

Long cases: it is nearly always time

The examiners are unusually consistent here. Time management is identified as the largest single factor in poor long case performance, and the failure pattern is specific. Candidates spend too long on the first two or three cases, write expansively about the ones they feel confident on, and leave inadequate answers for the rest. The Board's advice, repeated over many sittings, is that adequate answers to all six cases will pass where two or three excellent answers and three poor ones will not.

The other repeated observations are almost stylistic. Bullet points and short sentences score better than flowing prose. Repeating the same fact in different sections earns nothing extra. Pertinent negatives are expected, so if you diagnose a malignancy the examiners want to see you state the absence of spread to the sites it usually spreads to. And if you cannot name the diagnosis, describing the abnormality anyway can still gather marks that make the difference between passing and failing the component.

The oral: where the failures are behavioural as much as cognitive

The examiners' reports on the oral component read, at times, like a study of human behaviour under stress, and they are oddly comforting for it.

Candidates ask for the clinical history before they have looked at the image. They look at the examiner's face for a clue rather than at the film. They scroll repeatedly through the same series, which the Board once described, with evident sympathy, as using the mouse wheel to relieve stress rather than to find anything. They talk continuously without pausing, which sounds like confidence and actually denies the examiner any opening to guide them towards the marks. They mumble, and an examiner who cannot hear an answer cannot credit it. They request cross-sectional imaging before extracting what the plain film was shown to them for.

None of those is a knowledge deficit. All of them are rehearsable, and none of them will be rehearsed by reading.

On knowledge, the Board has been strikingly consistent for a decade: anatomy is frequently weak, plain film interpretation is weak, particularly chest and abdomen, and the clinical dimension, presentation and onward management, is thin. Paediatric and neuroimaging recur as difficult areas. Ultrasound is habitually overlooked in favour of CT or MRI even where it would be the better next test. And every modality shown is a separate scoring opportunity, which means a candidate who hurries past the plain film to get to the CT has thrown away marks that were sitting there.

Breadth, and why subspecialisation is a liability

The examination tests the whole of general radiology. Your working week very likely does not. A trainee or fellow deep in MSK or neuro reporting is becoming genuinely expert in one dialect while the examination will ask for fluency in all of them, including the paediatric radiographs and contrast studies they have not touched in eighteen months. Candidates writing about this afterwards put it starkly: two films you can make nothing of may not be recoverable however well the rest goes.

This is a coverage problem wearing the costume of a knowledge problem. The candidate is not less able than their colleagues. They have been specialising, which the job rewards and the examination does not.

The avoidable one: not knowing the software

Across many years the Board has complained, in increasingly plain language, that candidates arrive without having used the demonstration platform. It notes that this puts them at a disadvantage, wastes their own examination time and generates entirely avoidable stress. One report observed that candidates were not using the full resolution available to them, which almost certainly explains some of the complaints about image quality, and which means marks were lost to not having enlarged the image.

Under the current arrangements there is no opportunity to practise on the system immediately before the written components at the venue, as there once was. The demonstration site is therefore the only rehearsal available, and it is free.

It is a genuinely irritating way to fail.

The part that is not your fault

The College's own reports document a consistent gap between UK and global candidates. In one sitting UK candidates passed at 52% against 34% for global candidates. In another, 83% against 63%. In another, 74% against 54%. This is not inference from a third party; it is in the Board's published reports, sitting after sitting.

The Board has also been candid about why, in a way that is worth quoting to yourself if you trained outside the United Kingdom. It has stated that the examination is designed for UK trainees, that pathologies which are common in the UK will appear even where they are uncommon elsewhere, and that candidates are expected to be familiar with the diagnostic and management pathways of the National Health Service. None of that is knowledge of radiology. All of it is examinable in practice, and a candidate trained elsewhere is being tested on it without ever having been taught it.

Beyond the examination hall, a cohort study of 1,860 UK trainees found group-level differences in FRCR outcomes associated with ethnicity, age, gender and socioeconomic background, including markers such as parental education and free school meal eligibility. Its authors concluded these most likely reflect advantage and disadvantage accumulated across the whole of medical training rather than anything specific to the examination. The College has acknowledged that first-attempt pass rates for trainees of Black, Black British, Caribbean or African ethnicity have run substantially below those of their white counterparts. An earlier analysis of 2B attempts found the significant division to be UK against non-UK, without an ethnicity effect among UK candidates in that cohort, so the mechanisms remain contested rather than settled.

I raise it for one reason. A candidate who fails will search for an explanation and will reach first for a defect in themselves. The evidence says forces are operating on these outcomes that no individual created and no amount of additional revision would have dissolved. That will not change a result. It may change what somebody concludes about themselves, which matters more.

Anxiety, which is a performance problem and not a character flaw

In the survey of 258 recent candidates, 63% said anxiety had reduced their performance in the oral examination. In the long cases it was 40%, and in the written rapid reporting component 30%. The authors concluded that reducing anxiety in the oral would improve how fair the examination is perceived to be, which is a notable finding in a study that otherwise reported candidates thought the examination valid and acceptable.

Sixty-three per cent is not a minority of nervous people. It is most of the room. And performance under pressure is rehearsable: the wider medical education evidence shows simulation with deliberate practice outperforming conventional preparation for skills executed under pressure, with the benefit greatest when the rehearsal conditions resemble the real ones. Which is a technical way of saying that the first time you feel that particular pressure should not be in the examination.

The illusion of knowing

There is a well-replicated finding that re-reading produces a sense of mastery substantially in excess of actual recall, and that learners judge their own retention poorly. Reviewing familiar material feels productive and achieves little. Retrieving under test conditions feels worse and works far better.

Most candidates who fail were not idle. A good number were working hard on material they had already covered, because covered material is pleasant to revise and unfamiliar material is not.

And then there is life

Dates arrive later than anyone would choose, compressing preparation into a window nobody selected. Clinical work does not pause, on-call continues, children still need collecting. A radiologist preparing in forty-minute fragments at the end of long days is not preparing badly. They are preparing in the only conditions available, and a good deal of the advice written for people with clear evenings is quietly useless to them.

What follows from all this

Compressed, candidates fail for five broad reasons, and only one of them is knowing less radiology than the next person.

They fail on breadth, because the job specialises and the examination does not. They fail on reporting technique, because scoring five marks a case against a structured scheme at pace is a distinct skill nobody formally teaches. They fail on time, particularly in the long cases, where the examiners identify it as the largest single factor. They fail on performance under pressure, which is rehearsable and mostly is not rehearsed. And they fail for structural reasons documented in the College's own published reports, which have nothing to do with their ability.

There is also a sixth, which is simply that they were preparing for the previous version of the examination. If you take one thing from this article, let it be that the short cases are not the old rapid reporting with a new name, and that advice about calling normals and hedging your bets belongs to an examination that no longer exists.

Read the examiners' reports. They are free, they are specific, they are written expressly to help you, and they tell you where the marks went. It remains slightly baffling to me how few candidates ever open them.

A failed attempt tells you something precise about where you were weaker than you believed. That information is unwelcome, expensive and genuinely useful, and it is not the same thing as a verdict on whether you belong in this profession.

This article has been about 2B, because that is where the examiners' reports are richest and where the format has just changed. Part 1 and 2A fail people for their own distinct reasons, and each deserves the same treatment, which I intend to give them in articles to follow.

Sources and further reading

Royal College of Radiologists. Examiners' reports (radiology), Final FRCR Part B Examination Board, sittings 2016 to present. Available on the RCR website. All pass rates, mark-scheme examples and areas of candidate weakness above are drawn from these reports.

Royal College of Radiologists. FRCR Part 2B (CR2B) examination format and scoring system. RCR website.

Yeung A, Booth TC, Jacob K, McCoubrie P, McKnight L. The FRCR 2B examination: a survey of candidate perceptions and experiences. Clinical Radiology 2011;66(5):412-419.

Pakpoor J, Al-Tawarah Y, Muthuswamy K, Taylor SA, et al. Differential outcomes at the Fellowship of the Royal College of Radiologists (FRCR) exams according to demographic and socioeconomic factors. Clinical Radiology 2024. PMID 39327208.

Hawtin KE, Williams HRT, McKnight L, Booth TC. Performance in the FRCR (UK) part 2B examination: analysis of factors associated with success. Clinical Radiology 2014;69(7):750-757.

Booth TC, Martins RDM, McKnight L, Courtney K, Malliwal R. The Fellowship of the Royal College of Radiologists (FRCR) examination: a review of the evidence. Clinical Radiology 2018;73(12):992-998.

Woolf K. Differential attainment in medical education and training. BMJ 2020;368:m339.

Dunlosky J, Rawson KA, Marsh EJ, Nathan MJ, Willingham DT. Improving students' learning with effective learning techniques. Psychological Science in the Public Interest 2013;14(1):4-58.

McGaghie WC, Issenberg SB, Cohen ER, Barsuk JH, Wayne DB. Does simulation-based medical education with deliberate practice yield better results than traditional clinical education? A meta-analytic comparative review of the evidence. Academic Medicine 2011;86(6):706-711.

Originally shared by Dr Koshy Jacob on LinkedIn. Follow the series there