Rheumatoid arthritis (RA) and osteoarthritis (OA) are frequently treated with nonsteroidal anti-inflammatory drugs (NSAIDs). Although NSAIDs are an effective therapy for the pain and inflammation of arthritis, they are associated with serious side effects, particularly ulceration, bleeding, and perforation of the gastrointestinal (GI) tract. In this study, 1826 OA or RA patients who either had been taking NSAIDS for > or =6 months or had been unable to tolerate continuous NSAID use because of adverse GI symptoms or suspected NSAID-related gastroduodenal lesions were examined endoscopically for gastroduodenal lesions and ulcers. At the same time, the patients were asked to rate the severity of any GI symptoms they had been experiencing. Of the total number of patients studied, 817 (44.7%) were OA patients with a mean (+/- SD) age of 55.8+/-12.9 years, and 1009 (55.3%) were RA patients with a mean age of 53.1+/-13.1 years. Clinically significant gastroduodenal lesions were found in 37.1% of patients (n = 678); of these, 24.0% (n = 439) had ulcers. Gastric ulcers were more frequent than duodenal ulcers (14.8% vs 10.2% of patients; P < 0.05), and most gastric ulcers (72.0%) were found in the antrum of the stomach. The prevalence of gastroduodenal ulcers increased with age (P < 0.001), duration of OA (P < 0.001), and duration of current NSAID use (P = 0.019). The prevalence of gastroduodenal ulcers in patients taking NSAIDs for <1 year was 13.8%, compared with a nearly twofold higher prevalence (25.9%) in patients taking NSAIDs for periods of > or =1 year and up to 15 years. The prevalence of gastric ulcers was 32.6% in patients with a history of gastric ulcer but only 13.5% in patients with no GI history (previous gastric ulcer, duodenal ulcer, or upper GI hemorrhage). No relationship was found between the prevalence of gastroduodenal ulcers and sex (men, 22.4%; women, 24.9%) or prevalence of gastroduodenal ulcers and type of arthritic disease (RA, 23.6%; OA, 24.5%). The prevalence of gastroduodenal ulcers increased with the severity of GI symptoms (P = 0.007). These results provide further endoscopic confirmation of the association between NSAID use and gastroduodenal lesions and ulcers and support the contention that safer treatment alternatives to conventional NSAIDs are required.