The potentially damaging gastric and duodenal effects of dipyrone, a nonnarcotic analgesic agent, were evaluated in three phases in comparison to placebo and paracetamol. Three groups of 12 healthy adult volunteers were treated in a double-blind study, according to a cross-over, randomization sequence, using the double-dummy technique, for two 15-day periods, with dipyrone 3 g/day and placebo (group I), dipyrone 1.5 g/day and placebo (group II), and dipyrone 1.5 g/day and paracetamol 1.5 g/day (group III). An esophagogastroduodenoscopy was performed at the beginning and end of each treatment period. In the first treatment group, grade-3 and 4 mucosal lesions were found after dipyrone administration (3 g/day) in 3 of 12 (25%) subjects (multiple antral erosions, gastric ulcer and duodenal ulcer, 1 case each), whereas grade-2 mucosal lesions (antral erosions) were detected in 1 of 12 cases (8%) after the corresponding placebo treatment. The difference between the two treatments, however, was not statistically significant (p > 0.05). Only in the gastric ulcer case were subjective symptoms reported (feeling of hunger). At the 1.5-g/day dose (groups II and III), dipyrone produced no gastroduodenal lesions, the endoscopic results showing no appreciable difference between dipyrone and either placebo (p = 0.54) or paracetamol (p = 0.99). No subjective symptoms were reported in any of these subjects. Dipyrone, administered for 2 weeks, has effects on the gastric and duodenal mucosa comparable to those of paracetamol and placebo, though noticeable damage is detectable at a dosage of 3 g/day.