Background: Hip and knee osteoarthritis are among the most prevalent and disabling conditions affecting mainly older adults, with a risk of undergoing a total hip or knee replacement for the end stage. Increasing recommendations of conservative treatments have been reported as the first-line strategy for the initial management of this condition. Objective: This review sought to investigate the effects of supervised exercises versus non-pharmacological conservative therapies to reduce pain and disability levels on patients with hip or knee osteoarthritis. Methods: Three databases (PubMed, Cochrane Library, and ScienceDirect) were systematically searched for randomized-controlled trials published between 01-01-2001 and 31-10-2022. PEDro scale was used to assess the methodological quality of the included studies. The PRISMA guidelines were applied for this review. Results: Twelve randomized-controlled trials of moderate-to-high quality were included in this review. The studies involved 1,049 participants with a mean age (SD) of 64 (6) years old. The duration of the intervention and follow-up varied from 2 weeks to 16 months. Supervised exercises for hip or knee osteoarthritis were significantly less effective in terms of pain reduction (SMD=-0.40 [95%CI 0.16, 0.65], p=0.001) compared to home-based exercises (active musculoskeletal therapies), but not in terms of disability reduction (SMD=-0.04[95%CI -0.43, 0.36], p=0.86). There was a non-significant difference of supervised exercises compared to passive musculoskeletal therapies in terms of disability (SMD=0.21[95%CI -0.09, 0.50], p=0.17), or pain reduction (SMD=-0.19; [95%CI -0.57, 0.19], p=0.33). Conclusion: Supervised exercises were found to be less effective in reducing pain, but not in disability reduction when compared to home-based exercises.