We investigated the association between number of children and Parkinson's disease (PD) in two independent studies. In a case-control study, we identified all subjects who developed PD in Olmsted County, MN, from 1976 through 1995, and matched them individually by age (+/-1 year) and sex to population controls (193 cases and 193 controls). The replication study was a population-based cohort study of 6,341 subjects from Rotterdam, the Netherlands (2,610 men). In the Olmsted County study, men who fathered at least one child had an increased risk of PD (unadjusted OR, 2.7; 95% CI, 1.2-6.1; P = 0.02), and the risk increased with increasing number of children. The findings in women were not significant. In the Rotterdam Study, the risk of PD increased significantly with increasing number of children in men (test for linear trend, unadjusted; P = 0.04), but not in women. The findings from both studies remained consistent in direction but reduced in magnitude of the association, and lost significance after simultaneous adjustment for education, cigarette smoking, alcohol consumption, and coffee consumption. The independent replication in two distinct populations and using different epidemiologic study designs may suggest a link between the number of children and PD restricted to men.