This review seeks to describe data which support the concept that TNF alpha has a pivotal role in the pathogenesis of RA. There is now compelling evidence from in vitro studies and in animal models that this is the case. The in vitro studies suggest that the direct pathogenic effects of TNF alpha are amplified by its potential to act as a potent paracrine molecule, by inducing other pro-inflammatory molecules such as IL-1, and GM-CSF. It has yet to be established whether similar paracine effects of TNF-alpha are observed in the collagen type II mouse, or in human TNF alpha transgenic mice, where (as discussed in this review), TNF alpha undoubtedly has a pathogenic involvement. During the ongoing inflammation in RA, TNF alpha (and other mediators?) induce the production of soluble TNF-R which can act as natural inhibitors. This process is, however, inadequate within the inflamed synovium, due to the fact that cytokine interactions are very local, through autocrine and paracrine mechanisms. In this environment there is a continuous competition between the interactions of surface or soluble receptor with TNF alpha. It may well be that soluble TNF-Rs play a more important role in the periphery where they act as an essential clearance mechanism for the ligand. If, as the available data suggest, TNF alpha does indeed have a pivotal role in the pathogenesis of RA, this needs to be verified by demonstrating that removal of TNF alpha locally in the synovial joint of RA patients has a significant effect on clinical parameters of disease, and the progression of the disease process itself.