As part of a larger epidemiological study [Neuro-Epidemiology Project South-Tyrol (NEPT)], we investigated the accuracy of a mail questionnaire for parkinsonism in two languages (German and Italian). We administered the instrument to 40 randomly selected subjects with parkinsonism (Italian-speaking, n = 20; German-speaking, n = 20), attending our Parkinson's disease clinic regularly. Each patient was matched by age, sex and language with a subject without parkinsonism residing in the same South-Tyrol Province in Northern Italy. Subjects free of parkinsonism were recruited randomly from two group practices collaborating in the NEPT study. A questionnaire containing nine symptom questions and two additional questions about the patient's diagnosis of parkinsonism and/or treatment was mailed to each subject's home. Forty subjects with parkinsonism and 36 without parkinsonism participated in the study. All nine symptom items showed significant differences between affected and unaffected individuals. A combination of any three questions yielded the best balance between sensitivity (95%) and specificity (89%). There were no differences between the German- and Italian-speaking groups. We demonstrated the usefulness of a simple questionnaire for validating the diagnosis of parkinsonism irrespective of the primary native language.