To investigate whether concurrent administration of misoprostol reduces the occurrence of serious upper gastrointestinal complications, such as perforation, gastric outlet obstruction, or bleeding, in patients with rheumatoid arthritis who are receiving nonsteroidal anti-inflammatory drugs (NSAIDs). 6-month randomized, double-blind, placebo-controlled trial. 664 clinical practices of family medicine, internal medicine, or rheumatology in the United States and Canada. 8843 men and women (mean age, 68 years) receiving continuous therapy with any of 10 specified NSAIDs for control of symptoms of rheumatoid arthritis. Patients were enrolled between July 1991 and August 1993. Patients were randomly assigned to receive 200 µg of misoprostol or placebo four times a day. Development of serious upper gastrointestinal complications detected by clinical symptoms or findings (not by scheduled endoscopy). Serious upper gastrointestinal complications were reduced by 40% (odds ratio, 0.598 [95% CI, 0.364 to 0.982; P = 0.049]) among patients receiving misoprostol (25 of 4404 patients) compared with those receiving placebo (42 of 4439 patients). During the first month, more patients receiving misoprostol (20%) than placebo (15%) withdrew from the study, primarily because of diarrhea and related problems (P < 0.001). Risk factors for serious upper gastrointestinal complications were increasing age, history of peptic ulcer or bleeding, and cardiovascular disease. Patients with all four risk factors would have a 9% risk for a major complication in 6 months. In older patients with rheumatoid arthritis, misoprostol reduced serious NSAID-induced upper gastrointestinal complications by 40% compared with placebo.