OBJECTIVE To compare the efficacy and safety of nabumetone and diclofenac sodium in the treatment of patients with rheumatoid arthritis (RA). METHODS A hundred and twenty-five patients with active RA were enrolled in a randomized, open-label, controlled, multicenter, 12-week study treating with nabumetone (1 000 mg/day) or diclofenac sodium (75 mg/day). Clinical assessments were performed before beginning therapy, and after 4, 8, and 12 weeks of treatment. Endoscopy was performed respectively before and 12 weeks after the medication in all patients. RESULTS The total efficacy rate of nabumetone was as high as that of diclofenac sodium (92.31% vs 88.52%, chi(2) = 1.114, P = 0.573). nabumetone treated patients showed significant improvement both in erythrocyte sedimentation rate (38.74 mm/1 h vs 25.56 mm/1 h, F = 14.005, P < 0.01 in nabumetone group, and 42.74 mm/1 h vs 34.36 mm/1 h, F = 3.811, P > 0.05, in diclofenac sodium group respectively) and C-reactive protein (2.08 mg/dl vs 1.21 mg/dl, F = 6.495, P < 0.05 in nabumetone group; and 3.29 mg/dl vs 2.31 mg/dl, F = 2.968, P > 0.05 in diclofenac sodium group respectively). The variety and incidence of gastrointestinal (GI) symptoms attributed to nabumetone was comparable to that of diclofenac sodium. The mean score of endoscopic lesions in the nabumetone group was significantly lower in the end of the treatment than those of their base line scoring (P < 0.01), and there was no significant change with regard to the diclofenac sodium group. The incidence of ulcer disease associated with nabumetone was 0% and 2.70% with diclofenac sodium. CONCLUSIONS nabumetone is as effective as diclofenac sodium in controlling the signs and symptoms of RA. However, GI lesions under the endoscope in the nabumetone group seemed to be less severe than that of the diclofenac sodium group.