We tested the hypothesis that prostate-specific antigen (PSA) levels would be higher among prostate cancer deaths as compared with controls over time in the 25-year follow-up of the Multiple Risk Factor Intervention Trial of participants ages 35-57 at entry.