The authors first briefly review how the concept of COX-2 selectivity was brought to light, then tested against the known gastrotoxicity ranking of currently used NSAIDs, from the old classics to the most recent. One truly selective COX-2 agent--celecoxib--is now being marketed in an ever increasing number of countries. So far it seems to keep its main promises, i.e. high--albeit not total--safety regarding gastrointestinal adverse effects, and undisturbed platelet function. Association with warfarin drugs seems to raise no problems, but one should still be wary of possible renal side-effects. Efficacy, at least as assessed in osteoarthritis and rheumatoid patients, appears satisfactory. However, treatment of intense inflammatory crises, such as gout or ankylosing spondylitis, has not been assessed, as yet. Another COX-2 agent--rofecoxib--is on the brink of being released. Its even more potent COX-2 selectivity raises new issues. What about some COX-1 activity that several authors detected in rheumatic synovitis? On the other hand, in particular circumstances, organs such as the stomach, the kidney and small blood vessels, seem to have their homeostasis partly controlled by COX-2 mechanisms also. These questions should be answered soon, whilst clinical experience with the COX-2 agent builds up.