To investigate the association between idiopathic Parkinson disease (IPD) and reduced frequency of prior cigarette smoking, the authors compared the 29-year follow-up mortality rates and IPD incidence rates of men who were either cigarette smokers or nonsmokers at the time of enrollment in the Honolulu Heart Study (1965-1968). Based on IPD cases detected up to June 30, 1994, the age-adjusted incidence rate in smokers was less than half that in nonsmokers: 34.4 versus 94.2 cases per 100,000 person-years of pre-illness follow-up, respectively. When data were stratified by 5-year age group, lower IPD incidence in smokers was observed at all ages between 50 and 90 years. Age-specific mortality trends for smokers and nonsmokers with and without IPD suggested that increased mortality in IPD patients was mostly associated with IPD itself and not with smoking. The slight excess mortality in smokers without IPD, versus nonsmokers without IPD, appeared insufficient to account for the "missing" incident IPD cases in smokers. These IPD incidence and mortality data are not highly consistent with the "selective mortality" hypothesis, which attributes reduced prior smoking frequency, typically reported by persons with IPD, to accelerated mortality in undiagnosed IPD-affected persons who smoke. The "protective" association of cigarette smoking with IPD occurrence may thus be real, suggesting the need for further study of biologic mechanisms of protection.