From 1967 to 1979, 118 incident cases of idiopathic Parkinson's disease (IPD) and 236 age- and sex-matched controls from Rochester, MN, were identified. Medical records on patients and controls for 40 years preceding the diagnosis of IPD were reviewed. The relative risk (RR) for ever-smoked and IPD was not significantly different from unity (RR = 0.7, 95% confidence interval = 0.4 to 1.2). The mean age at diagnosis of IPD was significantly younger (p = 0.007) in the ever-smokers (68.8 years) compared with never-smokers (73.8 years), although this needs to be interpreted cautiously. It is concluded that smoking had no effect on the development of IPD. Within 5 years after the index date, a new diagnosis of dementia was made more often in cases than in controls (p = 0.01). Relative risk of IPD significantly increased when prior diagnosis of psychoneurosis and psychosomatic illness had been made.