The Rotterdam Criteria: The Two-of-Three Rule, Explained

Rotterdam is a city, not a test

Nothing is measured “by Rotterdam.” The criteria are named after the place a consensus workshop was held — a meeting sponsored by ESHRE and ASRM in Rotterdam, the Netherlands, in 2003, published the following year as a revised consensus statement.

Before it, the working definition came from a 1990 National Institutes of Health conference, which required both androgen excess and chronic anovulation. The Rotterdam group revised that because, in their words, the syndrome “encompasses a broader spectrum of signs and symptoms of ovarian dysfunction than those defined by the original diagnostic criteria.”

So when a clinic says it is using Rotterdam, it means the two-of-three rule below — and it means your ovaries are allowed to be part of the picture rather than irrelevant to it.

The rule: two of three, plus one thing everyone forgets

An adult meets the criteria when two of these three are present.

Criterion	2003 wording	How it is applied now
Oligo- or anovulation	Infrequent or absent ovulation	Cycle length under 21 or over 35 days, or fewer than 8 cycles a year, once you are more than 3 years past your first period
Hyperandrogenism	Clinical signs or raised androgen levels	Hirsutism, acne or female pattern hair loss, or a raised total or free testosterone measured by mass spectrometry
Polycystic ovarian morphology	≥12 follicles measuring 2–9 mm, and/or ovarian volume >10 mL	≥20 follicles in at least one ovary on a modern transvaginal scan, or ovarian volume ≥10 mL — or a raised AMH instead of a scan

The forgotten fourth requirement is exclusion. Rotterdam is explicit that PCOS remains a syndrome and that no single feature is sufficient on its own — other causes producing the same picture have to be ruled out first. In practice that means thyroid function, prolactin and 17-hydroxyprogesterone as a minimum, with targeted testing for Cushing’s syndrome or an androgen-secreting tumour where the picture suggests it.

Why Rotterdam was argued about for a decade

Widening the definition widens the population. That is not a criticism — it is arithmetic — but it is the whole of the controversy.

A 2016 systematic review and meta-analysis of 24 studies measured the gap directly: the pooled prevalence of PCOS was 6% (95% CI 5–8%) under the NIH definition and 10% (95% CI 8–13%) under Rotterdam. Same women, same planet, two different rules — and roughly two-thirds more people meet the broader one.

In between sat the Androgen Excess Society, which in 2006 published a position statement arguing that the condition should be considered first and foremost a disorder of androgen excess. Their rule keeps hyperandrogenism compulsory and allows either irregular ovulation or polycystic ovaries as the second feature. Its pooled prevalence in the same meta-analysis was also 10% (95% CI 7–13%).

Definition	Requires	Ultrasound used?	Pooled prevalence
NIH 1990	Androgen excess and anovulation	No	6% (5–8%)
Rotterdam 2003	Any two of three	Yes, as one of the three	10% (8–13%)
AE-PCOS 2006	Androgen excess plus anovulation or polycystic ovaries	Yes, as an alternative second feature	10% (7–13%)

The argument was settled by adoption rather than by proof. Rotterdam is what the 2023 international guideline uses, and that guideline is now applied in 196 countries.

The four phenotypes Rotterdam created

Two of three can be met in four different ways, and those four combinations became the four phenotypes. This is the part of Rotterdam that actually matters to you, because the label is identical across all four while the presentation is not.

Phenotype	Features present	Counts under NIH 1990?	Counts under AE-PCOS 2006?
A	Androgen excess + anovulation + polycystic ovaries	Yes	Yes
B	Androgen excess + anovulation	Yes	Yes
C	Androgen excess + polycystic ovaries (cycles regular)	No	Yes
D	Anovulation + polycystic ovaries (no androgen excess)	No	No

Phenotype D is the one the whole argument was about. It is the only combination that no earlier definition recognised, and reviews of the phenotype literature consistently describe the hyperandrogenic phenotypes as carrying more metabolic disturbance than the non-hyperandrogenic one — though that literature is observational and heavily affected by referral bias, since the women in clinic studies are the ones sick enough to have been referred.

The practical use of the table: ask which two criteria you met, not just whether you met two. That answer is what separates the four types from each other, and it is usually sitting in your notes already.

What changed between 2003 and now

Rotterdam has been refined twice without being replaced, both times because the measuring instruments improved rather than because the biology moved.

The follicle count went up. Ultrasound resolution in 2003 could not see what a modern transvaginal probe sees, so the original threshold of 12 follicles began flagging ovaries that were entirely ordinary. On current equipment the bar is 20 or more in at least one ovary.

The scan became optional. The 2023 international guideline added an AMH blood test as an alternative to ultrasound for defining polycystic ovarian morphology in adults — its single biggest change to the diagnostic pathway.

And the scan stopped being needed at all in many cases. If you have irregular cycles and hyperandrogenism, you already have two of three — the guideline says an ultrasound is not necessary, and neither is an AMH. Requesting one anyway adds cost and waiting time without changing the answer.

Where Rotterdam does not apply

Adolescents. Within eight years of a first period, both irregular cycles and hyperandrogenism are required, and the ovarian criterion is not used at all — no ultrasound, no AMH. Multi-follicular ovaries are normal at that life stage, so the third leg of Rotterdam would flag most of a school year group.

Anyone on the combined pill. The pill produces a scheduled withdrawal bleed regardless of ovulation, raises sex hormone-binding globulin and suppresses androgen production. Two of the three criteria are therefore unreadable on it, and a washout period is needed before testing means anything.

Perimenopause. Cycles become irregular for reasons that have nothing to do with PCOS, and follicle counts fall with age. The criteria were not validated for this stage of life.

The name changed in 2026. Rotterdam did not.

In May 2026 a global consensus involving 56 academic, clinical and patient organisations renamed the condition polyendocrine metabolic ovarian syndrome, or PMOS, published in The Lancet. The reasoning was that “polycystic” describes cysts that were never cysts, and hides the endocrine and metabolic half of a condition affecting one in eight women.

The criteria were not part of that decision. The two-of-three rule, the thresholds, the exclusions and the four phenotypes all survive the rename intact — as does the word “Rotterdam,” since the workshop happened where it happened. If you want the background on the change itself, the full story is here; the current checklist in the form your clinician uses it is set out in the diagnosis criteria post.

Your next step

Ask for your diagnostic record, and read it against Table 1. Specifically: which two criteria were documented, whether the ovarian criterion came from a scan or an AMH, and whether TSH, prolactin and 17-hydroxyprogesterone were checked.

If the answer is thinner than you expected, that is common rather than alarming — and it is a concrete thing to raise at your next appointment. There is a script for that conversation, and the rest of the diagnosis section assumes you know which two you met.