How PCOS Is Diagnosed: The Rotterdam Criteria

If you have been suspecting PCOS for years and are tired of being told your cycles are "just irregular," you are not imagining the pattern. A PCOS diagnosis is not one test, it is a clinical pattern, and the rules for naming it changed in 2023. This post is the long version of what your clinician should be checking, and how to read the result they give you.

PCOS, or polycystic ovary syndrome, is the most common endocrine condition in people of reproductive age, affecting somewhere between 8 and 13 percent of that population depending on the criteria used.⁵ Most of the people I see in clinic have been to two or three other doctors before they get a definitive answer. The international guideline group behind the 2023 update found that the average delay from first symptoms to diagnosis is more than two years.¹ Most people see three or more clinicians before someone names what is happening. That delay is not a personal failing.

It is the consequence of a syndrome that hides in plain sight. Irregular cycles can be dismissed as normal variation, acne can be blamed on stress, and the lab numbers only fit together when someone is willing to look at all of them at once.

Why does a PCOS diagnosis take so long?

PCOS sits in the awkward space between gynaecology, endocrinology, dermatology, and metabolic medicine. None of those specialties owns the whole picture, which is part of why the diagnosis has historically been slow. The other part is that PCOS does not have a single confirmatory test. There is no equivalent of a positive pregnancy test or a fasting glucose above 7. Instead, the diagnosis depends on recognising a pattern across cycles, skin, labs, and imaging.

In 2023 a Monash-led international working group, endorsed by 39 professional societies, published the updated international evidence-based guideline for the assessment and management of PCOS.¹ This is now the global reference document. The full guideline is open access through the Monash Centre for Health Research and Implementation.⁷ The 2023 update kept the Rotterdam framework but tightened several of the thresholds. It also made one significant addition: anti-Mullerian hormone, or AMH, can now be used as an alternative to ultrasound for the third diagnostic feature in adults, when high-resolution imaging is not available.¹

That matters for you for two reasons. First, if you had a scan three or four years ago that was reported as "normal ovaries," the criterion at the time may no longer apply. Second, if your clinician has not heard of the 2023 changes, you may be assessed against thresholds that have been retired. I work through both of those scenarios further down.

How is PCOS diagnosed with the Rotterdam criteria?

The Rotterdam criteria were first published in 2003 by a joint ESHRE/ASRM consensus workshop and remain the backbone of every modern PCOS guideline.² The original framework requires two of three features:

1. Oligo-ovulation or anovulation (cycles that are irregular or absent).
2. Clinical or biochemical hyperandrogenism (signs of androgen excess, or elevated androgen levels on a blood test).
3. Polycystic ovarian morphology on ultrasound, now expanded by the 2023 guideline to include elevated AMH as an alternative.¹

You only need two of the three. So a person with irregular cycles and elevated testosterone meets the criteria even with completely normal-looking ovaries. A person with regular cycles, hirsutism, and a polycystic-appearing ultrasound also meets the criteria. The combinations matter, and they sort people into the four phenotypes I describe below.

The Androgen Excess and PCOS Society (AE-PCOS) published a slightly stricter version of the criteria in 2009, which requires hyperandrogenism as one of the two features, plus either ovulatory dysfunction or polycystic ovarian morphology.⁶ Under AE-PCOS, the "non-hyperandrogenic" phenotype (D, described below) does not count as PCOS. The 2023 international guideline kept Rotterdam, not AE-PCOS, so the four-phenotype model still stands.¹ Some endocrinology clinics will use AE-PCOS internally, which is one reason different doctors sometimes give the same person different answers.

Before any of the three features is allowed to count, other causes have to be excluded. That part is non-negotiable. PCOS is a diagnosis of exclusion, even though it is the most common cause of the pattern.

Feature 1: Oligo-ovulation or anovulation

In adults, this means cycles shorter than 21 days or longer than 35 days, or fewer than 8 cycles per year.¹ If you are within the first year after menarche, irregular cycles are normal and do not count. Between one and three years post-menarche, the threshold is wider: cycles less than 21 or more than 45 days qualify as abnormal. After perimenopause begins, the criterion does not apply at all.

Regular cycles do not exclude PCOS. About 15 to 20 percent of people with PCOS ovulate regularly, falling into what is called phenotype C, the "ovulatory PCOS" pattern, where the diagnosis hinges on androgens and morphology instead.⁴ If your cycles are dependable but your acne or hirsutism is not, do not let a clinician tell you the cycle alone rules PCOS out.

When ovulation status is unclear, your clinician can confirm it three ways: a mid-luteal progesterone level (drawn on day 21 of a 28-day cycle, or 7 days before the expected period; for a 35-day cycle, draw on day 28), serial follicular ultrasound through a cycle, or basal body temperature charting that shows a sustained luteal-phase rise. Mid-luteal progesterone above roughly 10 ng/mL (about 30 nmol/L) is consistent with ovulation having happened.

Feature 2: Clinical or biochemical hyperandrogenism

Clinical hyperandrogenism means visible signs of androgen excess. Hirsutism is the most specific: dark, terminal hair in a male-pattern distribution (upper lip, chin, jawline, chest, lower abdomen, inner thighs). It is graded with the modified Ferriman-Gallwey score, where a value of 4 to 6 or more (depending on ethnicity) is considered abnormal.¹ Persistent adult acne, especially along the jawline, and androgenic alopecia (thinning at the crown or widening of the part) also count.

Biochemical hyperandrogenism means an elevated total or free testosterone, free androgen index, or DHEAS on a blood test. The trick here is the assay. Standard immunoassays are unreliable at the low concentrations seen in women, and the 2023 guideline now recommends mass spectrometry for borderline values.¹ If you are on combined hormonal contraception, androgens are suppressed, so the test cannot be interpreted; the guideline asks for a wash-out of at least three months before drawing androgens.

Ethnicity matters here too. The Ferriman-Gallwey score systematically underestimates androgen excess in East and Southeast Asian patients, who have less body and facial hair at any given androgen level. If you are in this group and your skin is showing the pattern but your hirsutism score is low, ask for biochemical testing rather than relying on the visible score.

Feature 3: Polycystic ovarian morphology, or elevated AMH

This is the feature that changed the most in 2023. On a transvaginal ultrasound performed with a transducer of 8 MHz or higher, polycystic ovarian morphology is defined as 20 or more follicles per ovary, or an ovarian volume of 10 mL or more.¹ Older 5 MHz probes give lower counts; under those, the threshold used to be 12 or more follicles per ovary. If your scan was done on an older machine, the count it produced cannot be compared against the 2023 thresholds.

If a high-resolution scan is not available, the 2023 guideline now accepts elevated AMH as an alternative.¹ There is no single universal AMH cutoff, because the assays differ. Your lab should provide its own assay-specific reference range. As a rough orientation, AMH above 35 pmol/L (5 ng/mL) is commonly elevated in PCOS, but use the assay-specific threshold, not the round number.

Two important exclusions. First, ultrasound is not recommended for diagnosis within 8 years of menarche, because multifollicular ovaries are common and physiologically normal in adolescents.¹ Second, the word "cyst" in the report is misleading: what is being counted are antral follicles, which are normal small fluid-filled structures every ovary contains. They are not pathological. I expand on what to look for in the scan report in PCOS on Ultrasound: What Doctors Actually See.

What gets ruled out before a PCOS diagnosis?

PCOS shares its presentation with several other conditions, and the guideline requires all of them to be excluded.¹ In practice, this is a short list of blood tests:

• Thyroid function: TSH, with reflex free T4 if TSH is abnormal. Hypothyroidism causes oligomenorrhoea, weight gain, and fatigue and is easy to confuse with PCOS. See Thyroid, TSH, and Fertility.
• Prolactin: drawn in the morning, fasted, without recent breast exam or heavy exercise. Hyperprolactinaemia causes anovulation and galactorrhoea, sometimes points to a pituitary adenoma, and is treated differently from PCOS. See Prolactin and Fertility.
• 17-hydroxyprogesterone: early morning, follicular phase. A value above about 6 nmol/L (2 ng/mL) prompts an ACTH stimulation test to exclude non-classic congenital adrenal hyperplasia, a 21-hydroxylase deficiency that imitates PCOS.
• Cushing's syndrome screen: only if features suggest it (purple striae, central obesity disproportionate to BMI, easy bruising, proximal muscle weakness). Not routine.
• Androgen-secreting tumour screen: total testosterone above about 5 nmol/L (150 ng/dL), or symptoms that came on suddenly with virilisation (deepening voice, clitoromegaly, rapid hirsutism progression). Also not routine, but worth flagging.

If any of these come back abnormal, the diagnostic label changes. PCOS-like symptoms with elevated TSH means hypothyroidism first, then reassess once thyroid is replaced. PCOS-like symptoms with raised 17-OHP and a positive stim test means non-classic CAH, which is managed by endocrinology, not gynaecology. The label matters because the treatments diverge.

What are the four PCOS phenotypes, and why do they matter?

Once the Rotterdam features are combined, four phenotypes drop out:⁴

• Phenotype A: all three features. Anovulation plus hyperandrogenism plus polycystic ovarian morphology. The classic presentation. Highest insulin resistance and metabolic risk; most challenging fertility profile.
• Phenotype B: anovulation plus hyperandrogenism, normal ovarian morphology. Lab and skin-driven. Still high metabolic risk.
• Phenotype C: hyperandrogenism plus polycystic morphology, regular cycles. Sometimes called "ovulatory PCOS." Skin-driven and imaging-driven, but cycles are dependable. Moderate metabolic risk.
• Phenotype D: anovulation plus polycystic morphology, no hyperandrogenism. Lowest metabolic risk and the closest to "just irregular ovaries." Not recognised as PCOS under AE-PCOS criteria but is recognised under Rotterdam and the 2023 guideline.

This is not academic. Phenotypes A and B carry the highest prevalence of insulin resistance and respond more strongly to insulin-sensitising agents like metformin and inositol. Phenotypes A and D, the anovulatory ones, are the ones most likely to need letrozole or another ovulation-induction agent to conceive. If your clinician tells you your phenotype, that is a good sign they have actually thought about your case.

What if the PCOS criteria do not fit cleanly?

Real readers do not always match the textbook. A few specific situations come up often enough that they deserve their own callout.

Lean PCOS: about 20 to 30 percent of people with PCOS have a BMI under 25.⁴ The criteria are identical, but lean phenotypes get missed for years because the mental model of "obese, hirsute, anovulatory" persists in older training. I have written the full version of this elsewhere: Lean PCOS: When You Don't Match the Stereotype.

Adolescents: the guideline deliberately delays the diagnosis until at least 8 years post-menarche, because both follicle counts and cycle irregularity are physiologically high in this group.¹ Until then, the language is "at risk of PCOS" rather than a fixed diagnosis. That is intentional, and it protects young people from a label that may not apply to them in adulthood.

Perimenopause: Rotterdam was never validated for this group, and androgens, AMH, and cycle length all shift in this window. Diagnosis here relies on lifetime history rather than current snapshot.

Post-pill: combined hormonal contraception suppresses androgens, suppresses LH, and prevents anovulation. The criteria cannot be applied while on the pill. I will not put a Rotterdam label on someone within three months of stopping hormonal contraception, and ideally I want six months of natural cycles before I commit. Otherwise you risk a false-positive diagnosis from rebound or a false-negative diagnosis from residual suppression.

After significant weight change: bariatric surgery, large weight loss, or large weight gain all move androgens, AMH, and SHBG. A scan and a panel done in the middle of that change will not give a stable picture. Wait for weight to stabilise, ideally for six months, before re-running the workup.

What should you ask before accepting a PCOS diagnosis?

When the report comes back, the questions to ask your clinician are concrete:

1. Which of the three Rotterdam features do I meet, and how was each one measured?
2. Were thyroid, prolactin, and 17-OHP all checked to rule out other causes?
3. Was the ultrasound performed transvaginally with a high-frequency (≥ 8 MHz) probe?
4. If AMH was used instead of imaging, was it interpreted against the lab's own assay-specific cutoff?
5. Which phenotype (A, B, C, or D) am I, and what does that mean for my fertility and metabolic plan?
6. Were any of the exclusions positive? If so, what is the new working diagnosis?

A clinician who can answer all six is using the 2023 framework. One who cannot is using something older. Both can still help you, but you may need to do more of the chasing in the second case.

If the criteria are not met but you still have the pattern, you are not gaslit and you are not wrong. Some readers sit at borderline AMH, borderline FG score, and 31-day cycles, and they do not meet two of three on any given assessment. In that situation the answer is to repeat the workup in 6 to 12 months, watch the trajectory, and treat symptoms in the meantime rather than waiting for a label.

What's next

• If diagnosis is confirmed, the next two posts cover the labs and the imaging in detail: The PCOS Blood Test Panel and PCOS on Ultrasound.
• If you want to see the symptom pattern laid out before you book the appointment, read PCOS Symptoms: What to Look For Before You Test.
• If you have a normal BMI and are wondering whether the diagnosis still applies, read Lean PCOS: When You Don't Match the Stereotype.
• If you have been dismissed by a clinician and want a second read, see Getting a Second Opinion on Your Fertility Workup.
• If the diagnosis is confirmed and you are ready to move into treatment, the Medicated Cycles hub is the next step. Letrozole is first-line for ovulation induction in PCOS, and the hub covers what that looks like.

Sources

1. Teede HJ, Tay CT, Laven JJE, Dokras A, Moran LJ, Piltonen TT, et al. Recommendations from the 2023 International Evidence-Based Guideline for the Assessment and Management of Polycystic Ovary Syndrome. Fertility and Sterility 2023;120(4):767–793. https://doi.org/10.1016/j.fertnstert.2023.07.025
2. Rotterdam ESHRE/ASRM-Sponsored PCOS Consensus Workshop Group. Revised 2003 consensus on diagnostic criteria and long-term health risks related to polycystic ovary syndrome (PCOS). Human Reproduction 2004;19(1):41–47. https://doi.org/10.1093/humrep/deh098
3. Dewailly D, Lujan ME, Carmina E, Cedars MI, Laven J, Norman RJ, et al. Definition and significance of polycystic ovarian morphology: a task force report from the Androgen Excess and Polycystic Ovary Syndrome Society. Human Reproduction Update 2014;20(3):334–352. https://doi.org/10.1093/humupd/dmt061
4. Lizneva D, Suturina L, Walker W, Brakta S, Gavrilova-Jordan L, Azziz R. Criteria, prevalence, and phenotypes of polycystic ovary syndrome. Fertility and Sterility 2016;106(1):6–15. https://doi.org/10.1016/j.fertnstert.2016.05.003
5. Bozdag G, Mumusoglu S, Zengin D, Karabulut E, Yildiz BO. The prevalence and phenotypic features of polycystic ovary syndrome: a systematic review and meta-analysis. Human Reproduction 2016;31(12):2841–2855. https://doi.org/10.1093/humrep/dew218
6. Azziz R, Carmina E, Dewailly D, Diamanti-Kandarakis E, Escobar-Morreale HF, Futterweit W, et al. The Androgen Excess and PCOS Society criteria for the polycystic ovary syndrome: the complete task force report. Fertility and Sterility 2009;91(2):456–488. https://doi.org/10.1016/j.fertnstert.2008.06.035
7. Monash University Centre for Health Research and Implementation. 2023 International Evidence-based Guideline for the Assessment and Management of Polycystic Ovary Syndrome. Monash University, 2023. https://www.monash.edu/medicine/mchri/pcos/guideline