A 26-year follow-up study of 8,006 men enrolled in the Honolulu Heart Program examined the effect of cigarette smoking on the risk of developing idiopathic Parkinson's disease. Cases were identified through an ongoing search of hospital records and by the review of death certificates and medical records of local neurologists. Men who had smoked cigarettes at any time prior to study enrollment in 1965 had a reduced risk of developing idiopathic Parkinson's disease (relative risk = 0.39). Examination of smoking by pack-years revealed an apparent dose-response effect on the risk of idiopathic Parkinson's disease, but not on the age of onset. Coffee drinking was also associated with reduced risk, apparently because of its association with cigarette smoking. Although the detrimental health effects of cigarette smoking would far outweight any possible protective effect for smoking and Parkinson's disease, the association of smoking with apparent protection may contribute to understanding the underlying pathogenetic mechanisms.