Data on the familial aggregation of Parkinson's disease (PD) remain conflicting. We conducted a historical cohort study of 1,001 first-degree relatives of 162 probands with PD and of 851 relatives of 147 control probands representative of the population of Olmsted County, Minnesota (from 1976 through 1995). In addition, we studied 2,713 first-degree relatives of 411 probands with PD referred to the Mayo Clinic from 1996 through 2000 and 625 spouses of PD or control probands. Whenever possible, relatives were interviewed and screened for parkinsonism either directly or through a proxy, and those who screened positive were examined or a copy of their medical record was obtained to confirm the diagnosis (family study method). Thirty relatives of population-based PD probands, 18 relatives of population-based control probands, 65 relatives of referral PD probands, and 8 spouses had developed PD. Combining population-based with referral PD probands and population-based control probands with spouses, the relative risk was modestly increased (relative risk, 1.71; 95% confidence interval, 1.11-2.64). Relatives of probands with younger onset (</=66 years; first tertile) had a more significantly increased risk (relative risk, 2.62; 95% confidence interval, 1.66-4.15), whereas relatives of probands with later onset had no increased risk. The relative risk decreased when the incidence of all types of parkinsonism was considered.