The prevalence estimates for primary dystonia range from two to 50 cases per million for early-onset dystonia and from 30 to 7320 cases per million for late-onset dystonia. From analysis of methodological information from 14 selected studies, we concluded that all studies on the basis of treatment settings or record-linkage systems, and two population-based surveys were probably flawed by incomplete ascertainment; the third population-based study provided the largest prevalence for late-onset dystonia but probably overestimated the prevalence of the disorder. Age and ethnic differences among study populations further biased comparisons of estimates. On the basis of methodologically more robust service-based studies and the likely percentage of underdiagnosis in a given area, more accurate prevalence estimates may be 111 per million for early-onset dystonia in Ashkenazi Jews from New York area, 600 per million for late-onset dystonia in northern England, and 3000 per million for late-onset dystonia in the Italian population over age 50 years.