It is suspected that male hormones are associated with the risk of prostate cancer. To test this hypothesis, we conducted a nested case-control study in a cohort of 6860 Japanese-American men examined from 1971 to 1975. At the time of examination, a single blood specimen was obtained, and the serum was frozen. After a surveillance period of more than 20 years, 141 tissue-confirmed incident cases of prostate cancer were identified, and their stored sera and those of 141 matched controls were assayed for total testosterone, free testosterone, dihydrotestosterone, 3-alpha-androstanediol glucuronide, androsterone glucuronide, and androstenedione. Odds ratios for prostate cancer, based on quartiles of serum hormone levels, were determined using conditional logistic regression methods. The odds ratios for the highest quartiles were 1.37 (95% confidence interval, 0.73-2.55) for 3-alpha-androstanediol glucuronide and 1.24 (95% confidence interval, 0.62-2.47) for androstenedione, but none of the differences was statistically significant. The results were unremarkable for the other four hormonal measurements. In addition, the patients and controls were compared by hormonal ratios (i.e., total testosterone:dihydrotestosterone), but the results were also unremarkable. The findings of this study indicate that none of these androgens is strongly associated with prostate cancer risk.