Useful research and diagnostic criteria for PCOS arose from a conference in 1990, whereby PCOS could be defined by: (i) clinical and/or biochemical hyperandrogenism, (ii) chronic anovulation, and (iii) exclusion of related disorders. The presence of "polycystic ovaries" was not included in this definition, which created significant concern since many women with PCOS have polycystic ovaries on ultrasound, and conversely women with this ovarian morphology have a higher prevalence of androgen excess and insulin resistance. More recently, at an expert meeting in 2003 in Rotterdam, it was recommended that PCOS be defined when at least two of the following three features were present, after exclusion of other aetiologies: (i) oligo- or anovulation, (ii) clinical and/or biochemical hyperandrogenism, or (iii) polycystic ovaries. These newer criteria effectively create additional phenotypes of PCOS (e.g. women with hyperandrogenism and polycystic ovaries but normal ovulatory function, and women with ovulatory dysfunction and polycystic ovaries but no clinical or biochemical evidence of hyperandrogenism). It remains to be demonstrated whether these phenotypes actually represent patients with PCOS. Nonetheless, the trend towards the use of uniform diagnostic criteria in studies of PCOS will increase the comparability and potentially the value of published research.