Two-hour oral glucose tolerance test results from the Walnut Creek Contraceptive Drug Study, a 10-year prospective follow-up study on the metabolic and medical effects of the "pill," were analyzed to determine the contribution of six progestins and synthetic estrogen to an observed decrease in glucose tolerance. Glucose tolerance data and specific oral contraceptive formulation data were obtained for 354 oral contraceptive users and 1732 nonusers. Multiple linear regression analyses, using age, body mass index, and specific progestin and estrogen dosages as independent variables, were undertaken to define dose-response relationships between the contraceptive steroids and fasting, 1-hr and 2-hr glucose values. The results of these analyses were consistent with the observation that the mean 1- and 2-hr values were higher in users of the "pill." The 16% prevalence of impaired glucose tolerance in pill users was substantially higher than that of 8% in the nonusers. Examination of the regression coefficients suggested that one progestin (norgestrel) was more metabolically "potent" than the others. The remaining progestins appeared to have either a modest or no metabolic effect, in terms of 1- and 2-hr glucose values. Synthetic estrogens were shown not to be related to measures of glucose tolerance.