BACKGROUND: Insulin resistance is a central feature of polycystic ovary syndrome (PCOS), and hyperinsulinemia contributes to anovulation, oligo or amenorrhea, hyperandrogenism and infertility in women with PCOS. The use of insulin sensitizers, such as metformin or thiazolidinedione, in PCOS is becoming increasingly accepted. The purpose of our study was to evaluate the therapeutic effects of metformin and rosiglitazone on the metabolic and reproductive derangement, and find parameters predicting their therapeutic efficacy in Korean PCOS women. METHODS: Sixty-two women with PCOS were recruited. The baseline characteristics, including BMI, glucose tolerance test, lipid profiles, sex hormones and hyperinsulinemic euglycemic clamp test, were assessed. After the administration of the insulin sensitizer (metformin 1.5g/day or rosiglitazone 4mg/day) for 3 months, the insulin sensitivity was reassessed. A drug response was defined as menstrual restoration or pregnancy. RESULTS: Of the 62 women with PCOS, 36 gained restored regular menstruation, and a further 5 conceived (a drug response rate of 66.7%). There were no significant clinical differences between responders and nonresponders. Twelve weeks after taking the drugs, the insulin sensitivity was significantly improved (M-value 4.7+/-0.2 vs. 5.5+/-0.4mg/kg/min, P<0.05), and the free testosterone levels(72.5+/-39.9 vs. 45.8 +/-3.8pmol/L, P<0.05) were significantly decreased, without significant weight reduction. CONCLUSION: Metformin and rosiglitazone restored menstruation in 66.1% of women with PCOS. Hyperandrogenemia and insulin sensitivity were significantly improved with the use of the two drugs. However, metabolic or hormonal markers for predicting the drug response could not be found.