We suggest that acute GVHD after marrow transplantation reflects (1) host injury due to the conditioning regimen followed by the production of inflammatory cytokines; (2) stimulation of mature donor T cells in the milieu of increased cell surface expression of leukocyte adhesion molecules and HLA molecules, followed by the autocrine production of IL- 2; and, finally, (3) recruitment and activation of additional mononuclear effector cells from donor marrow progenitors, which produce additional inflammatory cytokines, thus sustaining the response. The second step is critical for the amplification of the systemic inflammatory response, and it is absence in autologous, syngeneic, and T-cell-depleted transplants. These T cells may also contribute to the inflammatory cytokine network. Acute GVHD can occur in the absence of primary tissue injury in such settings as transfusion-related GVHD; however, it is likely that a greater HLA disparity between donor and host is required. We propose that inflammatory cytokine production is the final common pathway of acute GVHD. If this model is correct, control of cytokine dysregulation at any of several points should control GVHD. Further studies of GVHD and investigations of cytokine antagonists (eg, IL-4 or IL-10) or combinations of antagonists such as IL-1ra and soluble TNF receptor or pentoxifylline will allow us to determine the validity of this hypothesis.