Smoking is an important risk factor for many cancers, yet the relationship between smoking and prostate cancer remains uncertain. We investigated whether smoking affected the risk of prostate cancers within a large prospective cohort study of dietary and environmental cancer risk factors among men ages 50 to 71 upon enrollment in 1995-1996 (n = 283,312). Cox proportional hazards regression models with hazard ratios (HR) and 95% confidence intervals (95% CI) were adjusted for age, race, education, height, body mass index, physical activity, family history of prostate cancer, diabetes, self-reported health status, prostate-specific antigen testing, digital rectal exam, total energy, alpha-tocopherol, calcium, alpha-linolenic acid, selenium, red meat, fish, and tomato intake. There were 14,810 nonadvanced and 1,830 advanced incident prostate cancers identified through 2003, and 394 men died of their disease through 2005. Current smokers had a decreased risk of nonadvanced prostate cancer (HR, 0.82; 95% CI, 0.77-0.88), but an increased risk of fatal prostate cancer (HR, 1.69; 95% CI, 1.25-2.27). Former smoking was also associated with decreased risk of nonadvanced prostate cancers (HR, 0.89; 95% CI, 0.86-0.92), but not fatal prostate cancers (HR, 1.03; 95% CI, 0.83-1.27). There was no apparent association between smoking and advanced prostate cancer. A number of biologically plausible mechanisms could explain these results, including the direct effects of carcinogens in tobacco smoke and the resulting changes in sex hormone or growth factor profiles. These findings suggest that current and former smokers may be at decreased risk of being diagnosed with prostate cancer and current smokers are at an increased risk of dying from prostate cancer.