Although prostate cancer is a major disease, causal factors are only partially understood. We examined occupational risk factors for this disease in a large case control study among U.S. blacks and whites. The study included 981 new pathologically confirmed prostate cancer cases (479 blacks and 502 whites) diagnosed between 1986 and 1989, and 1,315 population controls (594 blacks and 721 whites) who resided in Atlanta, Detroit, and 10 countries in New Jersey, covered by population-based cancer registries. Information on occupation, including a lifetime work history, was collected by in-person interview. No clear patterns of risk were found for U.S. whites versus blacks, nor for white-collar versus blue-collar jobs. Farming was related to prostate cancer (OR = 2.17; 95% CI = 1.18-3.98). Risk was restricted, however, to short-term workers and workers in crop production. Risk was not limited to those farming after 1950, when widespread use of pesticides started. Risks increased with increasing years of employment in firefighting (chi 2trend, p = 0.02) and power plant operations (chi 2trend, p = 0.03), and were elevated among long-term railroad line-haulers (OR = 5.85; 95% CI = 1.25-27.4); jobs with potential polycyclic aromatic hydrocarbon (PAH) exposures. Risk was elevated among athletes (OR = 5.38; 95% CI = 1.48-19.6). However, most of the cases were athletes before 1960, so the potential use of anabolic steroids was excluded. Although some clues about potential occupational associations were found, the overall results show that occupation is not a major determinant of prostate cancer risk.