Summary Objective Different exercise types may yield different outcomes in osteoarthritis (OA) subgroups. The objective was to directly compare effectiveness of two exercise programs for people with medial knee OA and co-morbid obesity. Design We performed a participant- and assessor-blinded randomized controlled trial. 128 people ≥50 years with medial knee OA and body mass index ≥30 kg/m2 were recruited from the community. Interventions were home-based non-weight bearing (NWB) quadriceps strengthening or weight bearing (WB) functional exercise for 12 weeks. Primary outcomes were change in overall knee pain (numeric rating scale, range 0–10) and difficulty with physical function (Western Ontario and McMaster Universities Osteoarthritis Index, 0–68) over 12 weeks. Secondary outcomes included other pain measures, physical function, quality-of-life, global changes, physical performance, and lower-limb muscle strength. Results 123 (96%) participants were retained. There was no evidence of a between-group difference in change in pain (mean difference 0.73 units (95% confidence intervals (0.05,1.50)) or function (2.80 units (−1.17,6.76)), with both groups reporting improvements. For secondary outcomes, the WB group had greater improvement in quality-of-life (−0.043 units (−0.085,-0.001)) and more participants reporting global improvement (overall: relative risk 1.40 (0.98,2.01); pain 1.47 (0.97,2.24); function 1.43 (1.04,1.98). Although adverse events were minor, more NWB group participants reported ≥1 adverse event (26/66 (39%) vs 14/62 (23%), p = 0.04). Conclusions Both exercise types similarly improved primary outcomes of pain and function and can be recommended for people with knee OA and obesity. WB exercise may be preferred given fewer adverse events and potential additional benefits on some secondary outcomes. Registration Prospectively registered (Australian New Zealand Clinical Trials Registry #12617001013358, 14/7/2017).