This study involved a randomized parallel groups comparison of the effects of aspirin formulated as enteric-coated granules (25 subjects) or as buffered tablets (26 subjects) with that of a lactose placebo (5 subjects), on the gastric and duodenal mucosa, as determined by endoscopic examination 2 h after a fasting single 975-mg dose. A grading scale of 0 (no damage) to 4 (severe damage) was used. The granule formulation produced a statistically significant (p less than 0.05) lower severity (mean 0.40 +/- 0.58 vs. 3.00 +/- 0.94) and incidence (36% of subjects vs. 100%) of gastric lesions than the buffered aspirin formulation. None of the lesions produced by the granule formulation or the placebo was considered clinically significant by the blinded endoscopist, whereas 17 subjects on the buffered formulation (65%) had clinically meaningful stomach damage. The incidence of duodenal lesions was minimal and comparable for the two formulations.