An inverse association between cigarette smoking and idiopathic Parkinson's disease has been reported in several retrospective studies, but prospective evidence is available only for men. We assessed the association between the incidence of Parkinson's disease and smoking in two large prospective cohort studies comprising men and women. New cases of Parkinson's disease were identified in the Nurses' Health Study for 1976-1996, and in the Health Professionals Follow-up Study for 1986-1996. Smoking history was assessed at baseline and updated on subsequent biennial questionnaires. In women, the age-adjusted rate ratios (95% confidence intervals) for Parkinson's disease relative to never-smokers were 0.7 (0.5, 1.0) for past smokers, and 0.4 (0.2, 0.7) for current smokers. In men, the age-adjusted rate ratios for Parkinson's disease relative to never-smokers were 0.5 (0.4, 0.7) for past smokers, and 0.3 (0.1, 0.8) for current smokers. In both cohorts, the strength of the association decreased with time since quitting (among past smokers), increased with number of cigarettes per day (among current smokers), and increased with pack-years of smoking. These prospective findings confirm that an inverse association between smoking and the incidence of Parkinson's disease exists in both men and women.