Our previous studies have demonstrated that cultured human endothelial cells (EC) provide costimulation to PHA-activated CD4+ T cells, measured as augmentation of IL-2 synthesis, through a cell contact-department pathway. Here we show that fixed and living EC provide comparable degrees of costimulation to CD4+ T cell populations, indicating that EC costimulation does not depend upon active metabolism. EC achieve these effects in part by utilizing lymphocyte function-associated antigen-3 (LFA-3) to interact with T cell CD2 as shown by observations that EC augmentation of IL-2 is partially (50-70%) blocked by eight of eight mAb tested which recognize LFA-3; that purified phosphatidylinositol-linked LFA-3 (PI-LFA-3) can also provide costimulation to CD4+ T cells; and that there is a delay of the EC effect on CD4+ T cells which express low levels of CD2 compared to those which express high levels of CD2. However, three lines of evidence suggest that EC also utilize at least one additional ligand. First, there is incomplete replacement of the EC effect by PI-LFA-3 such that the costimulatory ability of EC combined with PI-LFA-3 is additive at all concentrations of PI-LFA-3 tested. Second, costimulation by PI-LFA-3, but not by EC, is fully inhibited by anti-CD2 or anti-LFA-3 mAb. Finally, costimulation by PI-LFA-3, but not by EC, is completely suppressed by cyclosporine A. We have not formally identified the second ligand but it does not appear to be intercellular adhesion molecule-1, vascular cell adhesion molecule-1, CD44, or B7/BB1.