How PCOS Is Actually Diagnosed, and Why AMH Was Added in 2023

The short answer#

Polycystic ovary syndrome is not confirmed by any single test. It is a pattern built from three possible features: irregular menstrual cycles, signs of excess androgen (clinical or on bloodwork), and polycystic ovarian morphology. Under the Rotterdam framework that the 2023 International Evidence-based Guideline carries forward, you meet the diagnosis as an adult when any two of those three are present and look-alike conditions have been ruled out. The headline change in 2023 was permission to use a serum anti-Mullerian hormone (AMH) level in place of a pelvic ultrasound for the third feature, and the same document also fixed an age boundary so the label is not stuck on healthy teenagers.

Key points#

• PCOS is diagnosed by counting features, not by one number. Two of three establish it in an adult.
• If irregular cycles and clear androgen excess are both present, that is already two features. No ovary imaging or AMH is needed.
• The third feature (ovarian morphology) is only a tiebreaker, used when one of the first two is missing or unclear.
• The 2023 guideline lets serum AMH substitute for ultrasound in adults, mainly to widen access where scanning is not available.
• Neither ultrasound nor AMH should be used to diagnose PCOS within eight years of a first period.
• AMH alone diagnoses nothing, and results depend heavily on which laboratory assay produced them.

Start with the counting rule, not the scan#

The most useful thing to understand about PCOS is its arithmetic. Because it is a syndrome, it is defined by a combination rather than by one abnormal result. The 2023 guideline, published in the Journal of Clinical Endocrinology and Metabolism, preserves the Rotterdam structure: any two of the three features are enough in an adult once mimics are excluded.

That structure carries a consequence people routinely overlook: if you already have irregular cycles together with clear signs of excess androgen, you have met two criteria and the diagnosis is complete. No ovarian imaging is required, and no ovarian bloodwork either. The guideline says so plainly. Practice Point 1.4.9 states that when irregular cycles and hyperandrogenism are both present, an ovarian ultrasound is not necessary for diagnosis, and Recommendation 1.5.2 extends the same reasoning to AMH: in that situation, the level is not needed.

So the third feature, polycystic ovarian morphology, is best pictured as a tiebreaker. It only matters when one of your first two is absent or uncertain. Reading the order the other way around, and treating a "polycystic-appearing" scan as if it alone means disease, is one of the most common mistakes in this area. A report saying your ovaries look polycystic does not, on its own, mean disease.

The age boundary that does most of the protecting#

Before getting to any test, the guideline's strongest safeguard is a rule about when not to test at all. In the years right after a first period, high follicle counts and high AMH are ordinary parts of a reproductive system that is still maturing. Interpreting them as disease during that window would mislabel healthy adolescents.

For that reason both measures are held back. Neither ovarian morphology on ultrasound nor serum AMH should be used to diagnose PCOS within eight years of menarche. Recommendation 1.4.6 advises against ultrasound in adolescents, and Recommendation 1.5.4 advises against AMH in the same group. In a teenager, then, the diagnosis narrows to the remaining two features, and both are required together: irregular cycles and hyperandrogenism.

The guideline also refuses to force a premature verdict. Under Practice Point 1.1.4, an adolescent with some features but not the full set can be flagged as at increased risk and reassessed at or before full reproductive maturity, roughly eight years after menarche. That is a deliberate trade: a scheduled second look instead of an early label.

What "polycystic ovarian morphology" really describes#

The phrase misleads in plain English. The follicles counted on ultrasound are not the painful cysts you are probably picturing. Morphology here is a counting threshold: 20 or more follicles measuring 2 to 9 millimeters in at least one ovary on a good-quality transvaginal scan. A high count is common in the general population, and more common in younger women, which is exactly why it cannot carry the diagnosis by itself.

Why a blood test was allowed to replace the scan#

Anti-Mullerian hormone is produced by the granulosa cells of small growing follicles. More such follicles means more hormone, so a blood AMH level tracks closely with the antral follicle count a sonographer would tally on screen. In PCOS, AMH tends to run roughly two to threefold higher than in unaffected women. That tight biological link is what let the guideline accept AMH as a stand-in for the ultrasound measure of morphology.

The driver was access, not novelty. A transvaginal ultrasound needs equipment, a trained operator, and a setting that many primary care and general practices simply do not have. A blood draw needs none of that. Recommendation 1.5.1 states that serum AMH could be used to define polycystic ovarian morphology in adults, which opens the diagnostic path to clinicians who cannot readily order imaging.

The guideline is also careful to stop the two tests from stacking. Practice Point 1.5.5 says either AMH or ultrasound may define morphology, but not both, precisely to limit overdiagnosis. Running two overlapping tests only raises the odds that a healthy person crosses a threshold by chance.

The fine print on AMH#

AMH is a surrogate, not a bypass around the framework. The evidence review behind the guideline is blunt that AMH used as a single marker for PCOS has poor sensitivity and specificity, and that AMH alone is not recommended. A high value diagnoses nothing on its own. It substitutes only for the morphology criterion, which must still be paired with irregular cycles or hyperandrogenism.

There is a laboratory catch as well. AMH assays are not standardized across manufacturers, so the same sample can return meaningfully different numbers on different platforms. Published cutoffs have ranged widely, roughly 3 to 10 nanograms per milliliter across studies, largely because the assays disagree. The guideline therefore tells laboratories to use population and assay-specific cutoffs rather than one universal line. Your number means little unless you also know which assay produced it and what threshold that assay uses.

Clearing the look-alikes#

Because these features overlap with other conditions, PCOS is a diagnosis of pattern plus exclusion. When androgen levels sit well above the laboratory reference range, the guideline directs clinicians to weigh other causes, including androgen-secreting tumors, non-classic congenital adrenal hyperplasia, and Cushing's syndrome. Thyroid disease and a raised prolactin are checked too, since either can disrupt your cycles. PCOS is what remains after those alternatives are addressed, which is the deeper reason no single result, AMH included, can carry the diagnosis alone.

This counting-and-excluding logic is also why PCOS sits comfortably in the metabolic medicine conversation, a research area of interest here, since the same reasoning about combining criteria and reading population-level cutoffs recurs across endocrine and cardiometabolic conditions.