If patients with parkinsonism were the only users of antiparkinsonian drugs and the drugs were utilized by the majority of the patients, the prevalence of the disease could be estimated from information on drug sales. The present study examines whether these prerequisites are fulfilled and whether the quality of the information on antiparkinsonian drug utilization in Sweden is sufficient for epidemiological use. Information on the use of antiparkinsonian drugs was collected from different registers of the Swedish National Corporation of Pharmacies. The quality of the data was good and made it possible to calculate the age-specific use of drugs by patients with parkinsonism. Levodopa derivatives (L-dopa + decarboxylase inhibitors) were found to be the best tracers for parkinsonism.