Epidemiology studies of parkinsonism employ a variety of techniques for unbiased sampling of populations. No current method permits mass screening of all subjects in a population for parkinsonism by movement disorders specialists. We developed and piloted a new approach to facilitate accurate and efficient screening of large populations for diagnosis of parkinsonism and provide data on sensitivity and specificity. We evaluated 2081 welders referred for medical-legal screening. Subjects were video taped using a standardized protocol, and videos were rated on the Unified Parkinson's Disease Rating Scale motor subsection 3 (UPDRS3). A "video rater" viewed video tapes and entered ratings through a web-based database. An "in-person" examiner performed a UPDRS3 examination in a randomly selected subgroup of 48 workers drawn from the 2081. We developed quantitative diagnostic criteria for parkinsonism that established minimum diagnostic thresholds based upon UPDRS3 scores and compared these criteria with diagnosis by an in-person examiner. Specificity of these criteria compared to in-person examination was 91-100% but sensitivity was 56%. A threshold UPDRS3 score greater than nine provided 100% sensitivity and 81% specificity. Liberal criteria identified 266 (13.1%) subjects with probable parkinsonism and 220 (10.8%) subjects with definite parkinsonism. Conservative criteria identified 260 (12.8%) with probable parkinsonism and 122 (6%) with definite parkinsonism. Our screening method permits rapid assessment of parkinsonian signs. An absolute UPDRS3 score greater than nine provided the best combination of sensitivity and specificity for the diagnosis of parkinsonism, while quantitative exam-based criteria for cardinal parkinsonian signs maximized specificity. Parkinsonism as diagnosed by our criteria was common in this group of welders.