Epidemiology of parkinsonism for 13 years (1967 to 1979) in Rochester, Minnesota is evaluated. For each patient, two age- and sex-matched controls with comparable medical follow-up (mean 40 years before diagnosis) were selected from the same community. Smoking habits, dementia, acute and chronic illnesses and mortality patterns in the two groups are compared. The annual incidence rate was 20.5/100,000, which is virtually unchanged from previous reports on Rochester population. Cigarette smoking habits of cases and the controls before the index date were not significantly different. Prior to diagnosis of Parkinson’s disease, dementia was diagnosed significantly more frequently among the patients than the controls (P = 0.048). After diagnosis of parkinsonism dementia emerged nearly three times more frequently in patients than among the controls. Mortality rate in the entire parkinsonian population was similar to that reported from the same community between 1935 and 1966. Only 50% of the patients were treated with levodopa. Among those treated with levodopa the mortality rate was comparable to their controls. Because there were no definite criteria for administering levodopa therapy, higher frequency of dementia and prolonged survival in those treated with this drug could be attributed to the bias of case selection for treatment.