The feasibility of determining the exact site and amount of drug-induced gastric bleeding was tested. Fourteen patients with rheumatoid arthritis received equivalent therapeutic doses of the antinflammatory drugs aspirin, 4 gm/day, and fenoprofen calcium, 2.4 gm/day, in randomized order for seven days. Acetaminophen was given for 14 days just prior to each of these periods. By fiberoptic gastroscopy, antral ulceration and acute mucosal lesions were found in seven patients following aspirin ingestion, in one taking fenoprofen, and in none taking acetaminophen. Fecal blood loss in four-day stool collections, quantitated by autologous chromium 51-labeled erythrocytes shed into the stool averaged 5.0 ml/day while taking aspirin, 2.2 ml/day while taking fenoprofen calcium, and 0.8 ml/day while taking acetaminophen. The mean blood loss was greater for those in whom gastric lesions developed while taking aspirin than for those in whom lesions did not develop. The short-term risk of erosive gastritis was greater for aspirin than fenoprofen.