Elevated insulin levels may explain part of the increased risk of endometrial cancer observed in obese postmenopausal women. Circulating sex hormones and fasting C-peptide levels were measured in sera obtained from 165 postmenopausal endometrial cancer cases accrued between June 1, 1987, and May 15, 1990, from hospitals in Chicago, Illinois; Hershey, Pennsylvania; Irvine and Long Beach, California; Minneapolis, Minnesota; and Winston-Salem, North Carolina, and 180 community and hysterectomy controls. Women with a personal history of diabetes were excluded. Among controls, C-peptide was positively correlated with body mass index (BMI) ((r = 0.44), waist-to-thigh circumference ratio ((r = 0.24), estrone ((r = 0.18), and estradiol ((r = 0.28) (albumin-bound (r = 0.45), and free (r = 0.37)) and negatively correlated with sex hormone-binding globulin (r = -0.48). In age-adjusted analyses, the odds ratios and 95% confidence intervals for tertiles of C-peptide and endometrial cancer were, from lowest to highest: 1.0 (reference), 0.78 (95% confidence interval (CI) 0.43-1.4), and 2.2 (95% CI 1.3-3.7). Further adjustment for BMI substantially attenuated the odds ratios for the highest tertile of C-peptide (odds ratio = 1.2, 95% CI 0.63-2.1), and adjustment for body mass index and other risk factors for endometrial cancer eliminated the association (odds ratio = 1.0, 95% CI 0.55-2.0). In contrast, adjustment for C-peptide had little influence on the magnitude of the positive associations between body mass index (odds ratio for highest vs. lowest tertile, without and with adjustment for C-peptide = 4.1 (95% CI 2.3-7.5) and 3.7 (95% CI 1.9-7.1), respectively) or several steroid hormones and endometrial cancer. These data are not consistent with the hypothesis that the effect of obesity on endometrial cancer risk is mediated through high insulin levels.