BACKGROUND AND AIMS: Non-pharmacological therapies are an important component of treatment for osteoarthritis (OA), but they may be under-used. This study examined the prevalence of self-reported use of common non-pharmacological therapies, as well as patient and physician-related predictors of use. METHODS: Subjects included 205 veterans who completed a survey regarding OA symptoms and treatments. Analyses examined the prevalence of use of three specific non-pharmacological therapies: exercise, physical therapy (PT), and dietary/herbal supplements. We also examined whether patient variables (demographics, clinical characteristics, and perceived helpfulness of non-pharmacological therapies) and physician characteristics (age, gender, race, and recommendation of non-pharmacological therapies) were associated with use of each therapy. RESULTS: Forty-six percent of subjects reported current use of exercise, 11% reported using PT, and 12%, dietary/herbal supplements. Patient demographic and clinical characteristics were generally poor predictors of use of non-pharmacological therapy. However, females were more likely to report exercising than males (p<0.05), and patients with greater disease severity were more likely to report current use of PT (p<0.001). Patients' perceived helpfulness of each therapy significantly predicted use (p<0.05). Physician demographic characteristics were not strong predictors of patients' use of therapy, but physician recommendation for exercise and PT predicted patients' use (p<0.05). CONCLUSIONS: Among this sample of veterans with OA, there was relatively low use of exercise, PT, and dietary/herbal supplements. Patients' perceptions of treatment helpfulness and physician recommendations strongly predicted use. These results signal the importance of interventions aimed at educating both patients and physicians about these therapies.