PCOS Rotterdam Criteria, Explained: How Diagnosis Actually Works

Polycystic ovary syndrome is the most common endocrine disorder in reproductive-age women, affecting somewhere between 6% and 13% of the population depending on which criteria are used. And yet PCOS diagnosis is genuinely difficult. Two patients with the same lab values can be diagnosed differently. Two clinicians looking at the same patient can disagree. Symptoms drift in and out of view depending on weight, stress, hormonal contraception, and life stage.

This post walks through the Rotterdam 2003 criteria in detail, what each component actually means, how the AE-PCOS Society and NIH 1990 criteria differ, and why PCOS diagnosis still requires clinical judgment rather than a single test.

Three competing criteria sets

Three diagnostic frameworks have been used historically:

Rotterdam is the most widely used internationally and is endorsed by the 2018 International PCOS Network guideline (Teede HJ, Misso ML, Costello MF et al., “Recommendations from the international evidence-based guideline for the assessment and management of polycystic ovary syndrome,” published 2018).

The reason for multiple criteria sets: each captures slightly different patient phenotypes. NIH 1990 is the strictest and risks under-diagnosis. Rotterdam is broader and risks over-diagnosis (especially in adolescents). AE-PCOS argues that hyperandrogenism is the defining feature and should be required.

The 2018 international guideline ultimately endorses Rotterdam with refinements — most importantly, requiring exclusion of other causes and tightening the polycystic morphology ultrasound criterion.

The three Rotterdam criteria, in detail

Criterion 1: Oligo-ovulation or anovulation

Operationally:

• Cycles longer than 35 days, or
• Fewer than 8 cycles per year, or
• Cycle length variability of 2–4 weeks (in adults; longer in adolescents within 3 years of menarche)

Anovulation is most directly demonstrated by:

• Mid-luteal progesterone less than 3 ng/mL (timing matters — needs to be 7 days before expected period)
• Lack of BBT shift across a tracked cycle
• Lack of LH surge

For tracking-based identification, see PCOS cycle tracking and anovulation and irregular cycles.

A common confusion: “regular cycles” do not rule out PCOS. About 20–30% of PCOS patients have clinically regular cycles but anovulatory or irregularly ovulatory cycles underneath. A serum progesterone is the cleanest test if cycles look normal but ovulation status is uncertain.

Criterion 2: Hyperandrogenism (clinical or biochemical)

Clinical hyperandrogenism:

• Hirsutism: Coarse, dark hair in male-pattern distribution (chin, upper lip, chest, abdomen, back, thighs). Quantified using the modified Ferriman-Gallwey score, where 4–6 or more is considered abnormal in most populations. Ethnic norms matter — South Asian and Mediterranean populations often have higher baseline scores.
• Acne: Persistent moderate-to-severe acne, especially adult-onset on the jawline and chin.
• Androgenic alopecia: Diffuse thinning at the crown or temporal recession.

Biochemical hyperandrogenism:

• Total testosterone: Above lab reference (typically >50 ng/dL or >1.7 nmol/L)
• Free testosterone: More sensitive than total testosterone; the 2018 guideline favors free testosterone or free androgen index calculated from total testosterone and SHBG
• Androstenedione: Elevated in some PCOS phenotypes; less specific
• DHEA-S: Mildly elevated in some PCOS; markedly elevated suggests adrenal source (consider congenital adrenal hyperplasia or adrenal tumor)

A patient with clear clinical hyperandrogenism does not need biochemical confirmation. A patient with normal clinical exam but borderline labs is harder to interpret.

Criterion 3: Polycystic ovarian morphology (PCOM)

The original Rotterdam 2003 definition:

• 12 or more follicles measuring 2–9 mm in either ovary, or
• Ovarian volume above 10 mL

The 2018 international guideline updated this to reflect modern ultrasound resolution:

• 20 or more follicles per ovary on high-frequency transvaginal ultrasound (8 MHz or higher), or
• Ovarian volume above 10 mL

The reason: with modern transducers, the original “12-follicle” threshold flagged a large fraction of normal women without PCOS as having PCOM. The 20-follicle threshold restores specificity.

Important caveats:

• PCOM does not equal PCOS. Up to 25% of women without any PCOS features have PCOM on ultrasound. PCOM is a finding that supports a PCOS diagnosis when combined with another criterion, not a stand-alone diagnostic test.
• In adolescents within 8 years of menarche, PCOM is not used as a criterion. The 2018 guideline explicitly excludes PCOM in this age group because of high baseline rates of PCOM in normal pubertal development.
• Ultrasound is not required if 2 of the other criteria are present. A patient with oligo-ovulation and clinical hyperandrogenism has PCOS by Rotterdam without needing an ultrasound.

The four PCOS phenotypes

Rotterdam’s “any 2 of 3” framework produces four phenotypes:

Phenotypes A and B carry the highest metabolic risk. Phenotype D is the most controversial — some experts argue it is not really PCOS but rather hypothalamic dysfunction with incidental PCOM. The AE-PCOS Society does not recognize phenotype D for this reason.

What must be excluded

Rotterdam criteria require ruling out other conditions that mimic PCOS. Standard exclusion workup:

• TSH: Hypothyroidism causes anovulation and weight changes
• Prolactin: Hyperprolactinemia causes anovulation
• 17-hydroxyprogesterone (early morning, follicular phase): Screens for non-classic congenital adrenal hyperplasia, which mimics PCOS closely
• 24-hour urinary cortisol or dexamethasone suppression test (if Cushing features present): Cushing syndrome causes irregular cycles and hyperandrogenism
• Gonadotropins (FSH, LH): High FSH suggests primary ovarian insufficiency; very low FSH/LH suggests hypothalamic amenorrhea
• Pregnancy test: Always rule out pregnancy as a cause of amenorrhea

If any of these are abnormal, the diagnosis is not PCOS until the alternative is investigated or ruled out.

What labs typically look like in PCOS

Common patterns (not diagnostic alone):

• LH:FSH ratio greater than 2:1 (classic but inconsistent — present in only ~50% of PCOS)
• Mildly elevated total or free testosterone
• Low SHBG (especially with insulin resistance)
• Mildly elevated DHEA-S
• Elevated AMH (often 2–4x normal — being explored as a future diagnostic marker but not yet in Rotterdam)
• Insulin resistance (fasting insulin elevated, HOMA-IR >2.5)
• Often impaired glucose tolerance on 2-hour OGTT

The 2018 international guideline recommends OGTT screening at diagnosis and every 1–3 years thereafter, given the high rate of glucose dysregulation.

Why diagnosis is hard

Five reasons PCOS diagnosis is genuinely difficult:

1. Hyperandrogenism is partly subjective. Hirsutism scoring varies between observers. Acne severity is qualitative. Hair-loss patterns can be confused with other causes.
2. Lab cutoffs are noisy. Testosterone assays vary between labs. SHBG affects interpretation. A single value can mislead.
3. PCOM thresholds depend on equipment. Ultrasound resolution has improved; old criteria over-call.
4. Symptoms wax and wane. Weight changes alter androgens and ovulation. Hormonal contraception suppresses both. Stress affects cycles. The same patient looks like PCOS one year and not the next.
5. The criteria are imperfect. Rotterdam is the consensus standard but has known over-diagnosis problems in adolescents and post-pill patients.

This is why the recommended approach is a thoughtful workup over time, not a single visit diagnosis. Tracking cycles for 6+ months, combined with appropriate labs and ultrasound, gives a much more reliable picture than a one-off test.

Implications of the diagnosis

A confirmed PCOS diagnosis changes care across several axes:

• Cycle management: Cyclic progesterone or hormonal contraception to protect the endometrium from unopposed estrogen
• Metabolic screening: OGTT, lipid panel, blood pressure
• Fertility planning: Earlier evaluation if not conceiving; ovulation induction (letrozole is first-line per 2018 guideline) if needed. See PCOS and fertility.
• Cardiovascular risk awareness: PCOS confers higher long-term metabolic and cardiovascular risk
• Mental health screening: Higher rates of depression and anxiety in PCOS

The bottom line

PCOS is diagnosed by Rotterdam 2003 criteria — any 2 of 3: oligo/anovulation, hyperandrogenism (clinical or biochemical), and polycystic ovarian morphology — after excluding other causes. The 2018 international guideline tightened the PCOM ultrasound criterion to 20+ follicles per ovary and explicitly excluded PCOM as a criterion in adolescents within 8 years of menarche. Track your cycles with the Period Calculator, bring detailed cycle data to your provider visit, and expect a workup rather than a single test. A solid diagnosis is worth the time it takes.