Non-steroidal anti-inflammatory drugs (NSAIDs) damage the gastric mucosa, and an important part of this effect is probably due to inhibition of prostaglandin synthesis. We have therefore studied various drugs for their ability to reduce prostaglandin and thromboxane formation by human isolated gastric mucosa. The overall relative potencies for inhibiting the endogenous production of PGE, 6-keto-PGF1 alpha and thromboxane B2 by mucosal pieces was generally: indomethacin = naproxen greater than ibuprofen greater than piroxicam; diflunisal, the prodrug sulindac, and the analgesic paracetamol usually had small or variable effects. This rank order was mainly similar to the inhibition of gastric microsomal PGE2 formation from exogenous arachidonic acid, the relative potencies being: indomethacin greater than naproxen greater than ibuprofen = piroxicam = diflunisal; again sulindac and paracetamol had little or no effect. The relative propensity of NSAIDs to cause gastric mucosal damage is controversial, but aspirin and indomethacin may be worst, and ibuprofen seems to be among the safest. Potency as an inhibitor of prostaglandin synthesis correlates better with the reported propensity for damage than does potency x dose. For reasons that are given in the discussion, this may indicate that gastric mucosal damage by NSAIDs with short or moderate half-lives is due largely to locally absorbed drug. Whereas inhibition of prostaglandin synthesis is probably the major cause of the damage, the simultaneous reduction of thromboxane formation might be advantageous for gastric mucosal integrity. Various implications arise from our hypotheses concerning the design of anti-inflammatory drugs.