Epidemiological evidence consistently indicates that aspirin or non-aspirin non-steroidal anti-inflammatory drug use is associated with the occurrence of gastric ulceration, gastric and ulcer bleeding, and ulcer death. Evidence on duodenal ulcer occurrence conflicts, possibly because of differences in study populations. A wide range of mechanisms could explain the occurrence of non steroidal-induced damage. These include inhibition of bicarbonate secretion, effects on mucus formation, and vascular actions. Not all effects are dependent on cyclo-oxygenase inhibition. Short-term studies in humans provide indications of likely therapeutic effects, but cannot demonstrate clinical efficiency. Although anti-secretory drugs and prostaglandins can protect patients against the development of duodenal or gastric ulcers, but not both, there is no clinical evidence which bears upon the critical issue of protection against complications.