The data presented in this review establish that cultured human endothelial cells have the capacity to present antigens to T cells and to do so in the context of costimulators that lead to effective T cell activation. These activities raise the possibility that venular ECs, at sites of delayed hypersensitivity reactions, could be the primary antigen-presenting cell to circulating memory T cells. This putative role of ECs can explain the rapid rate of initiation of memory responses because ECs are uniquely positioned to have physical access to the pool of circulating memory T cells. Studies also suggest that ECs may present alloantigens to circulating T cells in the context of transplantation, thereby initiating rejection reactions. Nevertheless, we repeat our caveat that these proposed antigen-presenting functions of ECs have not been established in vivo. Cytokine-mediated changes, particularly induction of adhesion molecules and synthesis of lymphocyte-activating cytokines, such as IL-8, provide ECs with the potential to recruit memory T cells to inflammatory sites independent of antigen specificity. Although these functions have also not been rigorously shown to occur in vivo, immunocytochemical studies of experimental and pathological tissues provide significant support for this proposal. Similar adhesive and activating functions of ECs may apply to preferential homing of pre-T cells to thymus and naive T cells to lymph node. We conclude by noting that the weight of evidence reviewed here supports the proposal that the vascular endothelium be considered an integral part of the in vivo immune system.