The association of cigarette smoking and mortality from prostate cancer was evaluated in 348,874 black and white men who were screened as part of the Multiple Risk Factor Intervention Trial (MRFIT). Current smoking status was assessed, serum cholesterol was measured, and demographics were recorded at screening; however, no information was collected on history of smoking, prostate screening, or diet. The vital status of each member of this cohort was ascertained through 1990. Death certificates were obtained from state health departments and coded by a trained nosologist. A total of 826 deaths due to prostate cancer occurred over an average of 16 years follow-up. The proportional hazards model was used to study the joint association of age, race, income, cigarette smoking, serum cholesterol level, and use of medication for diabetes mellitus on risk of death from prostate cancer. Statistically significant associations were observed with age (p < 0.01), cigarette smoking status (relative risk (RR) = 1.31, p < 0.01), black race (RR = 2.70, p < 0.01), and serum cholesterol (RR = 1.02 for 10 mg/dl higher cholesterol level, p < 0.05). Similar results were obtained when deaths that occurred during the first 5 years were excluded. Among cigarette smokers, there was some evidence of a dose-response relation (p = 0.20). The relative risk for those who reported that they smoked 1-25 cigarettes per day compared with nonsmokers was 1.21 (p = 0.04); the relative risk for those who reported smoking > or = 26 cigarettes per day compared with nonsmokers was 1.45 (p = 0.0003). These findings add to the limited evidence that cigarette smoking may be a risk factor for prostate cancer.