We tested the hypothesis that serum levels of testosterone (T), dihydrotestosterone (DHT), and the DHT metabolite 3 alpha,17 beta-androstanediol glucuronide are positively associated with the risk of prostate cancer. This nested case-control study was based on the cohort of men who donated blood to the Janus serum bank at Oslo University Hospital (Oslo, Norway) between 1973 and 1994. Cancer incidence was ascertained through linkage with the Norwegian Cancer Registry. The study included sera from 59 men who developed prostate cancer (cases) subsequent to blood donation and 180 men who were free of any diagnosed cancer (controls) in 1994 and were of similar age (+/- 1 year) and had similar blood storage time (+/- 6 months) to the cases. Neither T, DHT, nor the ratio T:DHT was associated with risk of developing prostate cancer. Compared to the bottom quartile, the odds ratio (OR) associated with the top quartile of T was 0.83 [95% confidence interval (CI), 0.36-1.93]; the OR for the top (compared to the bottom) quartile of DHT was 0.83 (95% CI, 0.36-1.94), and the equivalent OR for T:DHT was 1.31 (95% CI, 0.58-2.97). Similarly, 3 alpha,17 beta-androstanediol glucuronide showed no association with prostate cancer risk; the OR for the top (compared to the bottom) quartile was 1.10 (95% CI, 0.41-2.90). These results showed no association, positive or negative, between androgens measured in serum and the subsequent risk of developing prostate cancer.