We hypothesized that the administration of troglitazone (TGZ), an insulin-sensitizing agent of the thiazolidinedione class, would improve dyslipidemia associated with insulin resistance in polycystic ovary syndrome (PCOS). Three hundred and ninety-eight women with PCOS in a multicenter, double-blind trial were randomly assigned to 44 wk of treatment with: placebo or troglitazone (150, 300, or 600 mg/d). We examined the responses of circulating lipid and lipoproteins [total cholesterol, high density lipoprotein cholesterol (HDL-C), low density lipoprotein cholesterol (LDL-C), and triglycerides (TTG)] by treatment arm, and the influence of glycemic parameters on baseline levels and response to treatment. There was a high prevalence of abnormal baseline lipid parameters, as defined by National Cholesterol Education Program guidelines [total cholesterol, ≥200 mg/dl (35%); LDL-C, ≥130 mg/dl (31%); HDL-C, <35 mg/dl (15%); TTG, >200 mg/dl (16%)]. Baseline models showed that parameters of insulin action had poor predictive power on lipid parameters. There was no significant response of any of the circulating lipids to treatment with either placebo or one of the troglitazone arms (after correction for multiple analyses). There were favorable, but nonsignificant, trends in HDL-C (increase) and LDL-C (decrease) and a trend toward decreased circulating TTG in the 300- and 600-mg TGZ dose treatment arms, both in an intention to treat analysis (n = 375) and in study completers (44 wk; n = 152). There also was a minimal treatment effect noted when only subjects with abnormal baseline levels were examined, and responders differed little from nonresponders in terms of indices of insulin action. There is a substantial prevalence of clinically recognized dyslipidemia in the population of women with unrecognized PCOS without type 2 diabetes. Treatment with an insulin-sensitizing agent may have minimal impact on circulating lipids. Further surveillance and treatment of abnormal lipid levels may be necessary in these women.