OBJECTIVE: To assess the performance of a telephone-administered questionnaire suitable for use in 2-phase surveys in the detection of symptomatic hip and knee osteoarthritis (OA) cases. METHODS: A questionnaire was designed based on typical symptoms and self-reported OA diagnosis. Three groups of subjects were consecutively enrolled from rheumatology units at French university hospitals. The disease status, based on American College of Rheumatology criteria, was first confirmed by a rheumatologist. Subjects then completed the screening questionnaire administered by interviewers unaware of the diagnosis and the clinical examination results. Three screening strategies were evaluated. RESULTS: In all, 119 subjects with hip OA, 137 with knee OA, and 111 subjects with other rheumatic diseases with lower extremity symptoms were recruited. The highest sensitivity for both hip and knee OA was obtained with the strategy based on reporting the presence or absence of symptoms (>96%). The specificity of this strategy was low (42% for both joints). When taking into account the self-reported OA diagnosis, the sensitivity slightly decreased (>91%), and the specificity increased greatly, from 76% to 78%. The highest specificity was obtained with the third strategy, requiring a rheumatologist opinion (from 82% to 85%) at the expense of lower sensitivity (>90%). CONCLUSION: The questionnaire tested in this study is a simple, valid, and reliable instrument to screen symptomatic hip and knee OA. As such, it fails to reach complete accuracy and clinical examination and radiographs remain necessary for complete ascertainment procedure.