Inhibition of prostaglandin biosynthesis via inhibition of the fatty acid cyclooxygenase (COX) is the mechanism of action of non-steroidal antiinflammatory drugs (NSAIDs). This results in an inhibition of the inflammatory and pain-producing activities of prostaglandins at a site of tissue injury but also in inhibition of prostaglandin production in the gastrointestinal tract (GI) and platelets, i.e. sites where endogenous prostaglandins are possibly involved in control of physiological functions. The discovery of two COX isoenzymes, COX-1 and COX-2, and the detection of their separate function and regulation, has initiated the search for new and putatively more selective inhibitors of prostaglandin biosynthesis. Specifically, selective inhibitors of COX-2 were developed in order to improve the anti-inflammatory and analgetic specificity and potency of the compounds and to reduce side-effects in the GI tract. Available experimental and clinical data of selective COX-2 inhibitors, including flosulide, celecoxib or rofecoxib, suggest improved gastric tolerance as compared to conventional, non-selective NSAIDs. However, experimental evidence suggests that both, the analgetic and anti-inflammatory action of COXinhibitors, might also require inhibition of COX-1. COX-2-selective compounds at antiinflammatory doses might have other side-effects, and for example reduce vascular prostacyclin production. Evidence is accumulating that COX-2 might not only be considered as a putatively detrimental enzyme but rather a highly regulated enzyme that also contributes to tissue protection and is even constitutively expressed in healthy human stomach mucosa. This paper reviews some of these newer aspects of COX-2-selective inhibitors in clinical use and discusses their possible benefits and risks.