In 1990-1992, the authors investigated the association of total and free testosterone with the metabolic syndrome in postmenopausal US women not taking hormone replacement therapy (n=362) in a prevalent case-control study of carotid atherosclerosis. Free testosterone was estimated by using the free androgen index (FAI) (total testosterone/sex hormone-binding globulin ratio). The metabolic syndrome was defined as the presence of three or more of the following criteria: waist circumference > or =35 inches (88.9 cm), triglycerides > or =150 mg/dl, high density lipoprotein cholesterol <40 mg/dl, blood pressure >130/80 mmHg, fasting insulin > or =100 pmol/liter, or impaired glucose homeostasis (fasting glucose > or =110 mg/dl or diagnosed diabetes mellitus). FAI, but not total testosterone, was strongly associated with the metabolic syndrome. Compared with women in the lowest FAI quartile, those in the highest quartile had a fivefold greater odds of having the metabolic syndrome (odds ratio=5.38, 95% confidence interval: 2.70, 10.7) after adjustment for age, race, and carotid atherosclerosis status. In multivariate analyses, the three-component metabolic syndrome combinations that contained both hyperinsulinemia and hyperglycemia were most strongly associated with increased FAI (absolute increase=0.41-0.54 compared with that for women who did not have these combinations; all p's < 0.001). Higher FAI was associated with the hyperinsulinemia and hyperglycemia components of the metabolic syndrome. The role of androgens in glucose homeostasis in postmenopausal women requires further study.