OBJECTIVES--To compare the efficacy and gastroduodenal safety of a fixed-dose combination of diclofenac sodium 50 mg and misoprostol 200 micrograms twice daily with those of piroxicam 10 mg twice daily and naproxen 375 mg twice daily in patients with osteoarthritis. METHODS--A 4 week, randomised, double-blind, parallel-group, multicentre study was conducted in 643 patients with symptomatic osteoarthritis of the hip and/or knee, who required continuous non-steroidal anti-inflammatory drug therapy for 4 weeks and who were without significant upper gastrointestinal damage as confirmed by endoscopy. RESULTS--For patients who had pre- and post-treatment endoscopic examinations, gastroduodenal ulcers developed in 3 (1.5%) of 200 patients treated with diclofenac/misoprostol, 21 (10.3%) of 204 piroxicam-treated patients, and 17 (8.6%) of 198 patients receiving naproxen (Chi square = 13.771, p = 0.001). The improvement in the osteoarthritis severity index was greater in the diclofenac/misoprostol group than in the piroxicam group (p = 0.004). Changes in physician and patient global assessments showed no significant differences between treatment groups. The incidences of diarrhoea and abdominal pain were higher in the diclofenac/misoprostol group than in the piroxicam and naproxen groups. CONCLUSIONS--Diclofenac/misoprostol at twice daily dosing is associated with significantly fewer gastroduodenal ulcers than either piroxicam or naproxen. The efficacy of diclofenac/misoprostol in treating the signs and symptoms of osteoarthritis is at least comparable to that of piroxicam and naproxen.