A randomized double-blind trial was carried out in 20 patients with osteoarthrosis of the hip or knee to compare the efficacy and tolerance of treatment with diflunisal or naproxen. During the first 4 weeks, patients received either 250 mg diflunisal or 250 mg naproxen twice daily and this was increased by 250 mg daily in 5 patients on diflunisal and in 3 on naproxen for the second 4 weeks of the trial. The results of subjective assessments made before and at the end of Week 8 showed a trend in favour of diflunisal for improvement of symptoms, except for weight-bearing pain which was improved in only 1 patient in each group. More of the patients receiving diflunisal than naproxen considered treatment to have been satisfactory, and rated their response as equally as good as or better than previous medication. Diflunisal produced significantly high incidence of gastro-intestinal upsets, leading to the withdrawal of 2 patients at Week 4.