Smoking, obesity, alcohol, and physical activity can modulate the endocrine system, and therefore have been hypothesized to play a role in the etiology of prostate cancer. At baseline in 1982, 80 percent (n = 3,673) of the noninstitutionalized persons age 65+ in two rural Iowa (United States) counties were enrolled into the Iowa 65+ Rural Health Study. Follow-up for mortality was complete through 1993, and cancer experience was determined by linkage to the State Health Registry of Iowa cancer database for the years 1973-93. We analyzed data on 1,050 men aged 65 to 101 years (mean age 73.5) with a full interview in 1982 and with no documented cancer in the 10 years prior to baseline. Through 1993 (8,474 person-years of follow-up), there were 71 incident cases of prostate cancer. In a multivariate model, age, cigarette smoking (relative risk [RR] = 2.9 for currently smoking 20 or more cigarettes per day compared with never smoking; P trend = 0.009), greater body mass index (BMI) (wt/ht2) (RR = 1.7 for BMI > 27.8 kg/m2 compared with < 23.6; P trend = 0.1), and greater level of physical activity (RR = 1.9 for high activity level cf inactive; P trend = 0.05) were independent predictors of prostate cancer, and these associations were stronger for regional or disseminated disease at diagnosis. Percent change in BMI from age 50 to baseline was associated positively with risk (P trend = 0.01), and this association appeared to be stronger in heavier men. There were no data on diet. These findings suggest that smoking, overweight, and weight gain in later life are risk factors for prostate cancer and support a hormonal etiology; the positive association for physical activity confirms some previous reports, but remains without a credible biologic mechanism.