OBJECTIVE: To determine the long-term effectiveness (>/=6 months after treatment) of exercise therapy on pain, physical function, and patient global assessment of effectiveness in patients with osteoarthritis (OA) of the hip and/or knee. METHODS: We conducted an extensive literature search in PubMed, EMBase, CINAHL, SciSearch, PEDro, and the Cochrane Central Register of Controlled Trials. Both randomized clinical trials and controlled clinical trials on the long-term effectiveness of exercise therapy were included. The followup assessments were at least 6 months after treatment ended. Methodologic quality was independently assessed by 2 reviewers. Effect estimates were calculated and a best evidence synthesis was performed based on design, methodologic quality, and statistical significance of findings. RESULTS: Five high-quality and 6 low-quality randomized clinical trials were included. Strong evidence was found for no long-term effectiveness on pain and self-reported physical function, moderate evidence for long-term effectiveness on patient global assessment of effectiveness, and conflicting evidence for observed physical function. For exercise programs with additional booster sessions, moderate evidence was found for long-term effectiveness on pain, self-reported physical function, and observed physical function. CONCLUSION: The positive posttreatment effects of exercise therapy on pain and physical function in patients with OA of the hip and/or knee are not sustained in the long term. Long-term effectiveness was only found for patient global assessment of effectiveness. However, additional booster sessions after the treatment period positively influenced maintenance of beneficial posttreatment effects on pain and physical function in the long term.