We studied the influence of three sets of diagnostic criteria on the age- and sex-specific incidence of Parkinson's disease (PD) among residents of Olmsted County, Minnesota, for the period 1976 to 1990. Incidence cases of parkinsonism were detected using the medical records-linkage system of the Rochester Epidemiology Project. PD was separated from other types of parkinsonism using strict, intermediate, and broad criteria. We found 154 incident cases of PD using the strict criteria, 215 using the intermediate criteria, and 266 using the broad criteria. The incidence rate was consistently higher for men across all ages with all three sets of criteria; however, sex differences were more striking at older ages when using the broad criteria. In men above age 79 years, the incidence rate of PD declined with strict criteria, remained stable with intermediate criteria, and increased with broad criteria. The impact of diagnostic criteria on the age-specific incidence curve was less striking for women. When using the broad criteria, the risk of PD increased constantly with age in both sexes, suggesting that PD is an aging-related disease. Our findings suggest that the diagnostic criteria used to separate PD from other types of parkinsonism influence the magnitude of PD incidence and its distribution by age and sex.