OBJECTIVES: Osteoarthritis is prevalent and debilitating disease that affects nearly 21 million Americans. For clinicians to assess the impact of treatment strategies require reliable, valid and interpretable measures. We conducted a systematic review of WOMAC which is a disease specific instrument measuring HRQL of patients with osteoarthritis. The review considered the evaluative, discriminative properties and interpretability of the instrument. The last systematic review was published in 2001. METHODS: A structured literature review was performed to identify evidence related to measurement properties (reliability, internal consistency, responsiveness, and interpretability) of WOMAC using two databases (MEDLINE and HealthSTAR) from 1982 to December 2005. We included randomized controlled trials and observational studies. RESULTS: Of 307 and 137 papers, 14 met the inclusion criterion. Discriminative: We found high reliability (correlation coefficients) for domains of pain and physical function (>0.7), compared to the stiffness domain (>0.6). All domains had generally high construct validity in differentiating among patients with better/worse HRQL. The internal consistency (Cronbach's alpha) was high (>0.8). WOMAC generally had high construct validity. Responsiveness varied by domain and by intervention. WOMAC was more responsive in patients who underwent total hip or knee arthroplasty compared to other interventions. We found that the physical function domain was more responsive than other domains and this was more evident in patients with arthroplasty. The minimal score difference on 0-10 scale needed to detect clinical improvement (worsening) was 0.64(0.75), 1.03(0.67), and 0.29(0.72) for pain, physical function, and stiffness domain, respectively. CONCLUSIONS: WOMAC has good discriminative and evaluative properties. These properties provide an opportunity to measure new health technologies and other interventions in this cohort of patients with confidence. To benefit from the use of this measure as an interpretable one we need additional research in patients, particularly within clinical trials where other objective and interpretable measures are used.