BACKGROUND: Insulin resistance is considered a regular component of polycystic ovary syndrome (PCOS). However, several studies have failed to confirm insulin resistance in non-obese women with PCOS. The aim of the study was to identify whether insulin resistance is present in normal weight women with PCOS and the factors associated with insulin sensitivity. METHODS: Twenty-two normal weight (body mass index, BMI < 25 kg/m2) women with PCOS, and 16 age and BMI comparable control women with regular menstrual cycles were examined during their early follicular phase. The levels of serum hormones and lipids were measured. The visceral fat area was assessed by computed tomography at umbilical level. The standard 75g oral glucose tolerance test was performed to determine the glucose tolerance status. The insulin sensitivity was measured using the euglycemic hyperinsulinemic clamp technique (target glucose 90 mg/dL, insulin~1 mu/kg/min). RESULTS: The levels of free testosterone (1.9+/-0.6 pg/mL vs. 0.8+/-0.3 pg/mL, p<0.001), androstenedione (14.5+/-3.7 nmol/L vs. 8.8+/-1.3 nmol/L, p<0.001), LH (10.7+/-4.5 IU/L vs 4.6+/-4.8 IU/L, p<0.001) and FSH (5.8+/-1.7 IU/L vs. 4.2+/-2.4 IU/L, p<0.05) of the women with PCOS were significantly higher than those of the control subjects. The fasting plasma glucose (4.92+/-0.31 mmol/L vs. 4.42+/-0.61 mmol/L, p<0.01) and post glucose load plasma insulin (233.2+/-119.5pmol/L vs. 109.0+/-46.4 pmol/L, p<001) levels of women with PCOS were significantly higher than those of the control subjects. The glucose disposal rate (M value) was significantly lower in women with PCOS compared to the controls (5.3+/-1.2 mg/kg min vs. 6.7+/-1.6 mg/kg min, p<0.05), even after adjusting for age and BMI. There was no significant correlation of the M value with the anthropometric and a metabolic indices, and a multiple regression analysis of the M value showed no significant variables. CONCLUSION: Our non-obese women with PCOS showed significant insulin resistance compared to their age and BMI comparable control subjects, and-their insulin resistance may be an intrinsic defect not associated with other features, such as hyperandrogenemia or body fat distribution patterns.