PCOS Types: The Four Real Phenotypes, and How Diagnosis Works

What are the “types of PCOS” people are actually searching for?

Two entirely different classification systems answer to that phrase, and only one of them is something a clinician can write on a chart. The version most people find first — insulin-resistant, inflammatory, adrenal, and post-pill PCOS — comes from wellness content and symptom-quiz coaching, not from any clinical guideline. The 2023 international evidence-based guideline, built from a systematic review of the entire published literature and its 254 recommendations, does not use any of those four labels once. What it uses instead is four Rotterdam phenotypes, lettered A through D, each defined by which two of three diagnostic criteria a person meets. A full breakdown of what the popular four types get right and wrong lives here — this page’s job is different: it walks through the diagnostic process those four real phenotypes come out of, the tests in the order they actually happen, and what changed when the guideline was rewritten in 2023.

That’s not a dismissal of what people are actually looking for. Insulin resistance, inflammation, adrenal androgen excess, and pill-related symptom unmasking are all real, studied mechanisms — they just don’t sort anyone into four separate diagnostic boxes the way the popular framing implies. What does sort people into boxes, with an actual test behind each one, is the Rotterdam classification below.

How is a PCOS diagnosis actually made?

A PCOS diagnosis rests on two of three specific features, confirmed only after every other condition that produces the same picture has been ruled out. Both halves are required: a diagnosis made on one feature alone, or made without the exclusion bloods, is not a complete diagnosis, even when it gets written on a chart that way.

The three features — agreed at a 2003 consensus workshop in Rotterdam and still the backbone of the 2023 international guideline — are irregular or absent ovulation, clinical or biochemical excess androgens, and polycystic ovarian morphology on ultrasound, or a raised AMH in adults. Two of the three, present together, is a diagnosis. One alone is not, and which two you meet is exactly what determines your phenotype letter, covered in full below.

What are the Rotterdam criteria, and what did the 2023 guideline change?

The Rotterdam criteria were agreed by an ESHRE/ASRM consensus workshop in 2003 and widened an older 1990 NIH definition that had required both androgen excess and anovulation together. Rotterdam added ovarian appearance as a third, alternative criterion, so a person now qualifies with any two of three. A 2016 systematic review and meta-analysis of 24 studies measured exactly what that widening did: 6% of women met the older NIH definition, versus 10% under Rotterdam — roughly two-thirds more people, from the same underlying population, purely from a change in counting rules.

The 2023 international guideline kept Rotterdam’s two-of-three structure and made one substantive update: anti-Müllerian hormone (AMH) can now stand in for a pelvic ultrasound — in adults only. A meta-analysis of 82 studies commissioned for the guideline found AMH detected PCOS in the 68 adult studies it pooled with a sensitivity of 0.79 and a specificity of 0.87. In the 11 adolescent studies in the same analysis, both figures fell — to 0.66 sensitivity and 0.78 specificity — which is exactly why adolescent criteria don’t use AMH at all, and why a scan or a blood draw is never the whole story for anyone under 20. The same 82-study review also covers what AMH numbers do and don’t mean once you have a result.

One naming note, mentioned once because it matters and then set aside: in May 2026, a global consensus of more than 50 organisations renamed the condition polyendocrine metabolic ovarian syndrome, or PMOS. Nothing about the criteria, the phenotypes, or anything below changed with the name — the full story is here if you want it. This article uses PCOS because that’s still what most people search.

If you already have irregular cycles and clear androgen excess, neither an ultrasound nor AMH is required — you’ve already met two of three, and a scan at that point adds cost, not information. The full threshold list, cycle length by cycle length, sits on the diagnostic criteria page; what the ultrasound itself actually measures, including where a scan can miss the finding, is covered separately.

What are the four real PCOS phenotypes, and how do they differ in risk?

Two of three criteria can be satisfied in exactly four combinations, and those four combinations are the real phenotypes — lettered A through D, and this is what “types of PCOS” means in a medical chart. A prospective study of 1,212 women with PCOS found phenotype A in 48.2% of diagnoses, phenotype B in 30.7%, phenotype D in 11.4%, and phenotype C in just 9.7% — and, critically, the four groups did not carry equal metabolic risk.

Phenotype A carries the heaviest combined load of insulin resistance and androgen excess of the four, phenotype C shows the least metabolic disturbance at any body weight, and phenotype D is the only one where a normal-weight diagnosis does not, on its own, mean more insulin resistance than a person without PCOS at all. That ranking — not a wellness label — is what a clinician means by a phenotype carrying more or less long-term risk. If you don’t yet know which letter applies to you, the full phenotype breakdown, including which older definitions would and wouldn’t have recognised each one, is the place to start; ask your clinician to state the letter from your own chart rather than guessing from symptoms alone.

What does the popular “four types” framework get right — and get wrong?

Every mechanism behind the popular four types is real; none of the four is a diagnosable category the way the marketing implies. Insulin resistance is measurably concentrated in phenotypes A and B specifically, not spread evenly across all PCOS, so “insulin-resistant PCOS” describes a real gradient sitting inside two specific real phenotypes, not a fifth category layered on top of them. Adrenal androgen excess, tracked by a raised DHEAS on a blood test, shows up across every phenotype in roughly similar proportions rather than defining one exclusively. “Post-pill PCOS” describes something real that happens — a pre-existing anovulatory pattern becoming visible again once hormonal contraception stops suppressing it — but the pill cannot cause PCOS, only unmask it. And chronic low-grade inflammation is genuinely documented in PCOS, including in normal-weight women, but no lab test or cutoff sorts a subset of patients into an “inflammatory type” separate from the rest.

None of that makes the wellness framing worthless — it means it describes mechanisms operating inside your real phenotype, not an alternative to it. The point where it stops helping is when a quiz result replaces the two blood draws and the exclusion panel a real diagnosis requires. Lean PCOS is the clearest case: a normal BMI gets treated as reassurance that metabolic testing isn’t needed, when insulin resistance is present often enough in lean-phenotype patients that skipping the test misses it routinely.

Which tests should you expect, and in what order?

A complete work-up runs in a fixed sequence: cycle history first, because it costs nothing; bloods second, to score androgens and exclude look-alikes at the same time; imaging or AMH third, and only if the first two haven’t already settled the question.

Timing matters more than most people expect: getting the blood draw on the wrong cycle day is a common, avoidable reason a normal-looking result gets misread. The full blood test panel, listed test by test with what a high or low result usually means, expands on steps 2 and 3 above — including why a free testosterone reading can look normal while total testosterone is raised, or the reverse. Once insulin resistance is suspected rather than confirmed, the OGTT with insulin is the test most work-ups skip — a fasting glucose alone misses a hyperinsulinaemic pattern that a two-hour post-glucose insulin level catches.

What has to be ruled out before a PCOS diagnosis is confirmed?

Four conditions produce a picture close enough to PCOS that skipping the exclusion bloods is the single most common way a diagnosis goes wrong. Thyroid disease and high prolactin both stop ovulation and periods through mechanisms unrelated to PCOS, and both are checked with one blood draw each — TSH and prolactin. Non-classic congenital adrenal hyperplasia is the look-alike most often missed entirely: pooled data put it in roughly 4.2% of women presenting with androgen-excess symptoms, and it’s ruled out with a single morning 17-hydroxyprogesterone level — the full comparison, including the exact cutoffs, sits here. Where androgen excess is severe, rapid in onset, or accompanied by signs of virilisation, Cushing’s syndrome and an androgen-secreting tumour move onto the list too, though neither is routinely screened for in a standard work-up.

Rapid onset changes the urgency, not just the test list: hirsutism that arrived over months rather than years, or a deepening voice, warrants a faster, more urgent path than a routine PCOS work-up. Say so explicitly if it applies to you.

Skipping this exclusion step is also how PCOS gets over-diagnosed, not just under-diagnosed — the two different kinds of PCOS misdiagnosis go deeper into both directions than the summary above.

Why are the criteria different if you’re under 20?

Adolescent criteria drop a leg of the tripod on purpose: only two features are used — irregular cycles and hyperandrogenism, both required — and neither ultrasound nor AMH is used to diagnose PCOS within 8 years of a first period, under the guideline’s adolescent-specific recommendations. Multi-follicular ovaries and irregular cycles are both normal features of a maturing hormonal axis in the first few years after menarche, so a test built for adults would flag a large share of healthy teenagers as diseased. Where the picture is incomplete, the guideline supports an “at risk” label with symptom treatment and planned re-evaluation, rather than a diagnosis that might not hold.

This will not work for you if…

Every diagnostic framework has edges, and pretending otherwise is how people get told they’re an exception to a rule that was never designed to include them.

Your labs come back normal and you still have symptoms. PCOS with normal-looking bloods is a real, documented pattern — assay variation, wrong-day cycle timing, and phenotype C’s genuinely unremarkable metabolic profile can all produce a clean-looking panel in someone who still meets the clinical criteria on history and exam alone.

Your cycles are regular. PCOS with regular periods is phenotype C by definition — androgen excess plus polycystic ovaries, ovulation intact — and a clinician anchored on “irregular periods” as the entry symptom will miss it in 9.7% of cases by that route alone.

Your ultrasound looks unremarkable. PCOS without the follicle count the name implies is phenotype B, present in nearly a third of diagnoses, and a normal scan should never be read as ruling PCOS out on its own.

You’re on hormonal contraception right now. The pill produces a scheduled bleed regardless of ovulation and suppresses the androgen levels a diagnosis depends on. Two of the three criteria are unreadable while you’re on it; a washout period comes first.

Your symptoms started suddenly, or are severe and fast-moving. This entire framework assumes a gradual, years-long presentation. Rapid virilisation, a deepening voice, or hirsutism that progressed over weeks points toward a different and more urgent work-up — including rarer conditions that mimic PCOS closely enough to be missed for years if nobody asks the right question first.

Why does a diagnosis take so long, and what actually shortens it?

More than a third of women in one large survey reported a wait of over two years, and 47.1% saw three or more health professionals before a PCOS diagnosis was confirmed — and only 15.6% were satisfied with the information they received once it was. That’s self-selected, observational data from women recruited through support-group websites, not a population rate, but it matches what almost everyone reports. How long a diagnosis takes, and why breaks the delay down further.

The pattern behind that number is usually a fragmented history: the acne mentioned to a dermatologist, the periods to a GP, the weight to a nutritionist, with nobody holding all three pieces at once. Bringing a documented cycle log, a symptom timeline, and one direct request — “please check TSH, prolactin, 17-OH progesterone and androgens, and tell me which two Rotterdam criteria I meet” — closes most of that gap in a single visit. A full script for that conversation, organised by appointment type, covers what to ask next.

Your next step

Write down the start dates of your last six periods, when any androgen signs began and whether they came on gradually or fast, and whether you’re currently using hormonal contraception. Take that sheet to one appointment along with a single sentence: “I’d like to be assessed for PCOS — can we check TSH, prolactin, 17-hydroxyprogesterone and androgens, and confirm which two Rotterdam criteria I meet?”

That request names the exclusion bloods and the phenotype letter in one line — the two things most often skipped in a rushed visit. The rest of the diagnosis section picks up from whichever letter comes back.