Context: Polycystic ovary syndrome (PCOS) is defined most commonly according to the proceedings of an expert conference sponsored by the National Institutes of Health (NIH) in April 1990, which noted the disorder as having 1) hyperandrogenism and/or hyperandrogenemia, 2) oligoovulation, and 3) exclusion of known disorders. Alternatively, another expert conference held in Rotterdam in May 2003 defined PCOS, after the exclusion of related disorders, by two of the following three features: 1) oligoor anovulation, 2) clinical and/or biochemical signs of hyperandrogenism, or 3) polycystic ovaries. In essence, the Rotterdam 2003 expanded the NIH 1990 definition creating two new phenotypes: 1) ovulatory women with polycystic ovaries and hyperandrogenism, and 2) oligoanovulatory women with polycystic ovaries, but without hyperandrogenism. Objective: The objective of this study was to ascertain the validity of using the Rotterdam 2003 criteria rather than the NIH 1991 criteria for the diagnosis of PCOS. Intervention(s): Interventions included the use of the Rotterdam 2003 criteria for diagnosing PCOS and, in particular, the proposal to define two new phenotypes as PCOS. Positions: Available data suggest that hyperandrogenic ovulatory women with polycystic ovaries tend to have mild insulin resistance and mild evidence of ovarian dysfunction, although significantly less than women with anovulatory PCOS. However, whether these women will have an increased risk of infertility or metabolic complications, such as type 2 diabetes, remains to be determined. Alternatively, the risk of insulin resistance and long-term metabolic risks of oligoovulatory women with polycystic ovaries is even less well characterized and may be nonexistent. Conclusions: Because of the paucity of data on the two new phenotypes and their long-term implications and the potential negative impact on research, clinical practice, and patient insurability, the adoption of the Rotterdam 2003 criteria for defining PCOS should be considered premature. However, because polycystic ovaries are a frequent feature of PCOS, a modification of the NIH 1990 criteria is proposed. Additional research characterizing the phenotypes and associated morbidities of PCOS is urgently required. Copyright © 2006 by The Endocrine Society.