Summary Objective To evaluate the clinical effectiveness of manual physiotherapy and/or exercise physiotherapy in addition to usual care for patients with osteoarthritis (OA) of the hip or knee. Design In this 2 × 2 factorial randomized controlled trial, 206 adults (mean age 66 years) who met the American College of Rheumatology criteria for hip or knee OA were randomly allocated to receive manual physiotherapy ( n  = 54), multi-modal exercise physiotherapy ( n  = 51), combined exercise and manual physiotherapy ( n  = 50), or no trial physiotherapy ( n  = 51). The primary outcome was change in the Western Ontario and McMaster osteoarthritis index (WOMAC) after 1 year. Secondary outcomes included physical performance tests. Outcome assessors were blinded to group allocation. Results Of 206 participants recruited, 193 (93.2%) were retained at follow-up. Mean (SD) baseline WOMAC score was 100.8 (53.8) on a scale of 0–240. Intention to treat analysis showed adjusted reductions in WOMAC scores at 1 year compared with the usual care group of 28.5 (95% confidence interval (CI) 9.2–47.8) for usual care plus manual therapy, 16.4 (−3.2 to 35.9) for usual care plus exercise therapy, and 14.5 (−5.2 to 34.1) for usual care plus combined exercise therapy and manual therapy. There was an antagonistic interaction between exercise therapy and manual therapy ( P  = 0.027). Physical performance test outcomes favoured the exercise therapy group. Conclusions Manual physiotherapy provided benefits over usual care, that were sustained to 1 year. Exercise physiotherapy also provided physical performance benefits over usual care. There was no added benefit from a combination of the two therapies. Trial registration number Australian New Zealand Clinical Trials Registry ACTRN12608000130369.