In a study in the United States, Sweden, and Hungary, 335 cases of gastric bleeding without predisposing factors were compared with 670 control subjects, and 239 cases of duodenal bleeding were compared with 489 control subjects. For aspirin taken at least every other day during the week before the onset of bleeding (regular use), the relative risk of gastric bleeding was 4.4 (95% confidence interval [CI], 2.9 to 6.7); for occasional use, it was 3.3 (95% CI, 2.1 to 5.0). For ibuprofen, the corresponding estimates were 1.0 (95% CI, 0.4 to 2.6) and 1.1 (95% CI, 0.5 to 2.4). For naproxen, the estimate for regular use was 4.0 (95% CI, 1.5 to 11). The estimates for any use of piroxicam (crude estimate), indomethacin, and diclofenac during the week before onset were 18 (95% CI, 4.1 to 83), 1.6 (95% CI, 0.4 to 5.9), and 0.9 (95% CI, 0.2 to 4.2), respectively. The corresponding relative risks of duodenal bleeding were 7.1 (95% CI, 4.2 to 12) and 2.2 (95% CI, 1.3 to 3.7) for the regular and occasional use of aspirin, 2.4 (95% CI, 0.5 to 11) and 0.8 (95% CI, 0.3 to 2.0) for ibuprofen, 12 (95% CI, 2.8 to 54) and 9.9 (95% CI, 2.3 to 44) for naproxen, 17 (95% CI, 3.6 to 79) for any use of piroxicam (crude estimate), and 1.7 (95% CI, 0.2 to 14) for any use of indomethacin. There was a significant trend in the risk of gastric bleeding with increasing dose of regular aspirin use (p = 0.002). The relative risk estimates for the regular use of 325 mg or less were significantly elevated for both gastric and duodenal bleeding at 3.1 and 6.4, respectively.