We assessed the relationship between serum concentrations of estrogens, androgens, and sex hormone-binding globulin and risk of breast cancer among postmenopausal women. Study participants provided serum prior to breast biopsy or mastectomy in 3 hospitals in Grand Rapids, Michigan between 1977 and 1987. A total of 179 subjects with localized breast cancer were compared to 152 subjects with nonproliferative breast changes that have not been associated with elevated breast cancer risk. Increasing serum concentrations of estrone and estrone sulfate were associated with increases in breast cancer risk; the odds ratios (ORs) in the fourth quartiles compared to the first were 2.3 (95% confidence interval (CI) 1.1-4.6) for both (p-trend = 0.02 and 0.03, respectively). Estradiol and bioavailable estradiol concentrations were associated with nonstatistically significant increases in risk. Androstenediol levels were associated with risk (p-trend = 0.01); the OR in the fourth compared to the first quartile was 2.2 (95% CI 1.0-4.6). Testosterone, dehydroepiandrosterone and androstenedione levels were not associated with increased risk. Sex hormone-binding globulin was associated with a nonsignificant decrease in risk. Associations with estrone and estrone sulfate persisted after adjustment for androstenediol (ORs for fourth compared to first quartiles were 2.0 (95% CI 0.9-4.5) and 2.2 (95% CI 1.0-4.6), respectively (p-trend = 0.16 for both). The association with androstenediol was attenuated after adjustment for estrone (OR for fourth compared to first quartile was 1.6 (95% CI 0.7-3.6); p-trend = 0.13). Higher serum concentrations of estrogens were associated with increased breast cancer risk in postmenopausal women. Androgen levels were not independently associated with substantially increased risk.