Comparison of Parkinson's disease (PD) prevalence and incidence in various parts of the world is difficult because methods of case ascertainment, diagnostic criteria, classification, medical facilities, and age distribution of the populations vary broadly in different studies. We minimized these differences by adjusting available data to a single standard population. Using this we calculated age-adjusted rates for 27 regional populations and analyzed PD frequency from 45 communities. We conclude: (1) with the exception of China, Japan and Africa, which have the lowest prevalence ratios, the actual prevalence variation for PD is probably lower than previously reported in geographically diverse populations; (2) geographic variation is unlikely to be due exclusively to racial factors, and (3) environmental risk factors for PD might differ regionally.