Recently, the term of "possible" polycystic ovary syndrome (PCOS) has been used for defining cases in which biochemical evaluations are incomplete but clinical phenotypes are suggestive of PCOS. The aim of this study was, by using Rotterdam 2003 criteria, to detect possible PCOS cases and compare their characteristics and insulin sensitivity status with confirmed PCOS subjects. One-hundred-eighteen women who admitted with complaints and symptoms suggesting PCOS were included. Insulin sensitivity status of the cases was calculated with Homeostasis Model Assessment of Insulin Resistance (HOMA-IR). Cases fulfilling Rotterdam 2003 criteria were defined as confirmed PCOS, whereas indeterminate subjects as possible PCOS. Confirmed PCOS was detected in 70 (59.3%) and possible PCOS in 48 (40.7%) cases. Confirmed PCOS was most prevalent among subjects with hirsutism and menstrual dysfunction; 32 (80.0%) vs. 8 (20%), (p=0.000). Body mass index and HOMA-IR values did not differ between groups: confirmed PCOS versus possible PCOS; 25.46+/-5.55 kg/m(2) vs. 26.75+/-7.55 kg/m(2), 3.37+/-4.12 vs. 3.21+/-2.50, (p>0.05). Family history of type-2 diabetes mellifus was similar within both groups (p>0.05). Many PCOS patients seem to be undiagnosed due to inadherence to diagnostic work-up and/or to not fulfill Rotterdam 2003 criteria. These criteria may not be sufficient to cover the entire spectrum of PCOS.