Because the role of exercise in prostate cancer is unclear, we examined the relationship between leisure time physical activity and risk of prostate cancer in the Health Professionals Follow-up Study, a prospective cohort study of male health professionals in the United States. In 1986, 47,542 men 40-75 years of age and free of cancer responded to a mailed questionnaire that included an assessment of physical activity. The reported average time per week spent on each of a variety of nonoccupational activities was multiplied by its typical energy expenditure requirements expressed in metabolic equivalents (METs) and summed to yield a total weekly MET-hour score. We also examined MET-hours from all vigorous activities, defined as those requiring energy expenditures of six or more METs, and nonvigorous activities. From 1986 until January 31, 1994, we identified 1362 incident cases of total prostate cancer (excluding stage A1), 419 advanced (extraprostatic) cases, and 200 metastatic cases. No relationship with total or advanced prostate cancer was evident for total, vigorous, and nonvigorous physical activity. For metastatic prostate cancer, we also found no linear trends for these activities, but did observe a significantly lower risk in the highest category of vigorous activity (multivariate relative risk = 0.46; 95% confidence interval = 0.24-0.89 for > 25 versus 0 MET-hours), controlling for age, vasectomy, history of diabetes, height, smoking, and dietary factors. This highest category included 15% of the population and reflects at least 3 h/week of participation in vigorous activities. Differences in disease surveillance according to activity level could not account for our findings. The results from this cohort indicate that physical activity is unlikely to influence the incidence of total prostate cancer appreciably; however, the suggestion of a lower risk of metastatic prostate cancer in men engaging in high levels of vigorous activities warrants further study.