It is now well established that cyclooxygenase exists in two major isoforms, namely cyclooxygnease-1 (Cox-1) and cyclooxygenase-2 (Cox-2). The constitutive isoform, Cox-1, has been well characterized, and the sheep, mouse and human forms have been cloned (DeWitt and Smith, 1988; DeWitt et al., 1990; Yokoyama and Tanabe, 1989). It has been proposed that Cox-1 is involved in the maintenance of prostaglandin-mediated physiological functions, in particular cytoprotection in the stomach. In contrast, Cox-2, which has been extensively studied recently (O’Neill et al., 1994; Percival et al., 1994; Smith and Marnett, 1994; Herschman, 1994), is present in negligible amounts in normal stomach tissues (Kargman et al., data to be published). Cox-2 has been shown to be induced substantially in vivo under inflammatory conditions (Kennedy et al., 1993; Masferrer et al., 1994; Harrada et al., 1994). This has led to the hypothesis that Cox-1 and Cox-2 serve different physiological and pathological functions. Currently available non-steroidal anti-inflammatory drugs (NSAIDs) are non-selective cyclooxygenase inhibitors, which inhibit both Cox-1 and Cox-2 (Mead et al., 1993; Mitchell et al, 1994; Brideau et al, 1995). It is believed that gastrointestinal ulcerogenic effects commonly associated with these NSAIDs may be related to mechanism-based toxicity, i. e. via an inhibitory action on Cox-1 in the gastrointestinal tissues. On the other hand, if the anti-inflammatory activity of NSAIDs is predominantly due to the inhibition of Cox-2, selective inhibitors of Cox-2 may comprise a new class of therapeutic agents which possess similar anti-inflammatory effects to conventional NSAIDs, but with a substantially improved side effect profile. In this article, the pharmacology of a selective Cox-2 inhibitor, L-745,337 (5-methanesulfonamido-6-(2,4-difluorothiophenyl)-1 -indanone) in various in vitro and in vivo animal models is reviewed.