Hyperinsulinemia secondary to a poorly characterized disorder of insulin action is a feature of the polycystic ovary syndrome (PCO). However, controversy exists as to whether insulin resistance results from PCO or the obesity that is frequently associated with it. Thus, we determined in vivo insulin action on peripheral glucose utilization (M) and hepatic glucose production (HGP) with the euglycemic glucose-clamp technique in obese (n = 19) and nonobese (n = 10) PCO women and age- and body-composition-matched normal ovulatory women (n = 11 obese and n = 8 nonobese women). None had fasting hyperglycemia. Two obese PCO women had diabetes mellitus, established with an oral glucose tolerance test; no other women had impairment of glucose tolerance. However, the obese PCO women had significantly increased fasting and 2-h glucose levels after an oral glucose load and increased basal HGP compared with their body-composition-matched control group. There were statistically significant interactions between obesity and PCO in fasting glucose levels and basal HGP (P less than .05). Steady-state insulin levels of approximately 100 microU/ml were achieved during the clamp. Insulin-stimulated glucose utilization was significantly decreased in both PCO groups whether expressed per kilogram total weight (P less than .001) or per kilogram fat free mass (P less than .001) or when divided by the steady-state plasma insulin (l) level (M/l, P less than .001). There was residual HGP in 4 of 15 obese PCO, 0 of 11 obese normal, 2 of 10 nonobese PCO, and 0 of 8 nonobese normal women. The metabolic clearance rate of insulin did not differ in the four groups. We conclude that 1) PCO women have significant insulin resistance that is independent of obesity, changes in body composition, and impairment of glucose tolerance, 2) PCO and obesity have a synergistic deleterious effect on glucose tolerance, 3) hyperinsulinemia in PCO is not the result of decreased insulin clearance, and 4) PCO is associated with a unique disorder of insulin action.