The authors compared the lifetime occupational physical activity of 452 prostate cancer cases identified through the population-based Hawaii Tumor Registry and 899 population controls interviewed from 1981 to 1983. Each job reported was classified into one of five levels of physical activity using published sources. Among men aged 70 years or older, a negative association was found between prostate cancer risk and proportion of life spent in jobs involving only sedentary or light work. Compared with men never employed in such jobs, men who spent more than 54% of their life in these jobs had an odds ratio of 0.5 (95% confidence interval 0.3-0.9). This negative association was dose-dependent, consistent across ethnic groups, and unrelated to socioeconomic status, dietary risk factors, or job-related chemical exposures. The findings for younger men were less clear, but not inconsistent with those for older men. Surprisingly, no association was found with years spent in moderately active or very active jobs in either age group. Although inconclusive, these results suggest that physical activity may be positively associated with the risk of prostate cancer, but this association is likely to be weak and indirect.