Inflammation is a complex process involving numerous mediators. Because prostaglandins (PG) have been implicated as mediators in all stages of inflammation, inhibition of their synthesis provides the basis for the therapeutic effects of nonsteroidal antiinflammatory drugs (NSAID). Treatment with NSAID is usually accompanied by gastric side effects, attributed to interference with the formation of cytoprotective PG in gastric mucosa. An ideal NSAID should inhibit PG synthesis at the site(s) of inflammation but not in gastric mucosa. Experimental and clinical data support the view that this criterion has been met by etodolac, a structurally distinct NSAID. Thus, in rats and humans with rheumatoid arthritis, longterm daily administration of etodolac at effective antiinflammatory dosages (3 mg/kg in rats; 600 mg in humans) had no effect on PGF2 and prostacyclin levels in gastric mucosa. In contrast, significant decreases in gastric PG levels occurred with antiinflammatory doses of aspirin, naproxen, and piroxicam. Cyclooxygenase (COX), the pivotal enzyme in PG biosynthesis, exists in 2 isoforms: constitutive COX-1, which produces the PG required for maintenance of normal cell activity (e.g., gastric cytoprotection), and COX-2, which is induced in restricted tissue-specific fashion (e.g., by inflammatory stimuli). The antiinflammatory action of NSAID may result from inhibition of COX-2, whereas their gastric side effects may result in large part from inhibition of COX-1; thus, a preferred NSAID should inhibit COX-2 but not COX-1. Results show that etodolac has 10-fold selectivity for COX-2 and indicate that etodolac's pharmacotherapeutic efficacy can be explained by its demonstrably selective inhibition of COX-2, amplified by its favorable tissular pharmacokinetics. The sparing of COX-1 activity in gastric mucosa gives rise to etodolac's noted gastric tolerance.