The aim of the study was to compare the efficacy and the effects on the mucosa of the gastrointestinal tract (GIT) of nabumetone and diclofenac retard in patients with osteoarthritis (OA). An open, multicentre, randomised, comparative, endoscopy-blind parallel group study included 201 patients with nabumetone and 193 patients with diclofenac retard suffering from moderate to severe OA of the knee or hip joint. Twelve clinical efficacy variables were assessed and a portion of the population underwent gastroduodenoscopy. All patients exhibited significant improvement in pain severity and pain relief (p < 0.001 and p < 0.0001, respectively) but there were no differences between the groups for al the efficacy variables. Eleven per cent of patients on nabumetone and 1 with diclofenac underwent gastroduodenoscopy. The differences in the mucosal grade for the oesophagus, stomach and duodenum at baseline were not significant. In the oesophagus there were significantly less changes after treatment with nabumetone (p = 0.007) than with diclofenac; there were similar findings in the stomach (p < 0.001) but the difference in the duodenum was not significant. This study indicates that nabumetone and diclofenac retard have similar efficacy in the treatment of OA, but nabumetone has significantly fewer GIT side-effects.