A 4-week double-blind study compared the potential for 20 mg/day tenoxicam or 100 mg/day diclofenac sodium to induce gastropathy in 36 patients with joint disease and assessed the influence of gastric colonization by Helicobacter pylori. Endoscopic assessment at the end of 4 weeks indicated that the mucosa was normal in 79% of tenoxicam-treated patients and 59% of diclofenac-treated patients. Only 5% of patients in the tenoxicam group developed severe gastroduodenitis (> 11 haemorrhages or erosions) compared with 18% in the diclofenac group. Histological evaluation indicated that 58% and 47%, respectively, of tenoxicam-treated and diclofenac-treated patients retained normal mucosa after treatment. Diclofenac treatment was discontinued in two patients, due to a duodenal ulcer or severe erosive gastritis. Overall, 5/14 patients with moderate to severe colonization with Helicobacter pylori developed severe chronic active gastritis or ulceration, compared with the 1/22 patients in whom colonization was either absent or mild (P = 0.02). Tenoxicam and diclofenac did not show major differences in terms of gastrointestinal safety, although the trends favoured tenoxicam. The presence of severe colonization of the gastric mucosa with Helicobacter pylori appears to be an important factor for development of severe gastritis or ulceration.