Several lines of evidence suggest that sex hormones may be involved in the etiology of prostate cancer. We conducted a prospective nested case-control study to evaluate the relationships of serum androgens and estrogens to prostate cancer using serum collected at baseline for the Alpha-Tocopherol, Beta-Carotene Cancer Prevention Study. The 29,133 male smokers who participated in the trial were 50-69 years old at baseline. During 5-8 years of follow-up, 246 men were diagnosed with prostate cancer, and 116 of these were randomly selected for inclusion in the current study. For each case, two controls matched on age, date of blood collection, intervention group, and study center were selected. Hormones were measured in serum by RIA using standard procedures. None of the individual androgens or estrogens was significantly related to prostate cancer. These findings were unaltered by simultaneous evaluation of serum androgen and estrogen concentrations in multivariate models. These results do not support a strong relationship of serum androgens and estrogens with prostate cancer in smokers. Within-person variation in concentrations of some hormones may have contributed to the lack of significant associations.