Our objective was to investigate whether behavioral graded activity (BGA) has particular benefit in specific subgroups of osteoarthritis (OA) patients. Two hundred participants with OA of hip or knee, or both (clinical American College of Rheumatology, ACR, criteria) participated in a randomized clinical trial on the efficacy of BGA compared to treatment according to the Dutch physiotherapy guideline (usual care; UC). Changes in pain (Visual Analog Scale, VAS), physical functioning (Western Ontario and McMaster Universities Osteoarthritis Index, WOMAC, and McMaster Toronto Arthritis Questionnaire, MACTAR), and patient global assessment were compared for specific subgroups. Subgroups were assigned by the median split method and were analyzed using analysis of covariance. Beneficial effects of BGA were found for patients with a relatively low level of physical functioning (p?0.03). Furthermore, beneficial effects of BGA in patients with a low level of internal locus of control were marginally significant (p = .05). Patients with a relatively low level of physical functioning benefit more from BGA compared to UC. Compared to UC, BGA is the preferred treatment option in patients with a low level of physical functioning.