A total of 226 patients with osteoarthritis of the hip, knee or spine was entered into a general practice, multi-centre, crossover study of 500 mg naproxen twice daily and 400 mg ibuprofen 3-times daily. Patients were randomly allocated to receive either naproxen or ibuprofen first and were given each drug consecutively for 3 weeks with no washout periods. Both drugs significantly reduced from admission values the duration of inactivity stiffness, resting pain, movement pain, night pain, interference of the disease with daily activities and overall disease severity. Naproxen was superior to ibuprofen in relieving resting pain, movement pain, night pain and interference with daily activities. In the patients receiving ibuprofen first, all parameters showed a significant mean improvement after crossover to naproxen. Patients receiving naproxen first did not significantly improve further when crossed over to ibuprofen. Side-effects were reported by 64 patients. Forty-five patients had side-effects while taking naproxen and 30 while taking ibuprofen (11 patients had side-effects on both drugs). Most side-effects related to the gastro-intestinal tract. Many were mild and may not have been attributable to the treatment. One patient had a gastro-intestinal bleed while taking naproxen. Thirty-one patients (13.7%) failed to complete the study: 16 stopped while taking ibuprofen, and 15 while taking naproxen. Nine of these patients were lost for reasons not directly attributable to the treatment. Overall treatment preference significantly favoured naproxen.