A major sequela of immunotherapy with interleukin 2 (IL-2) is development of a vascular leak syndrome. The pathogenesis of this toxic effect is not known. We have examined pre- and post-treatment skin biopsies from 14 patients undergoing systemic administration of IL-2 for evidence of endothelial cell activation. Specifically, we have used the immunoperoxidase technique to detect the expression of three different activation antigens: endothelial-leukocyte adhesion molecule 1, detected with monoclonal antibody H4/18; intercellular adhesion molecule 1, detected with antibody RR1/1; and histocompatibility leukocyte antigen-DQ, detected with antibody Leu 10. Each of these antigens may be induced on cultured endothelial cells by various cytokines (although not by IL-2) and is expressed during endothelial cell activation in vivo at sites of delayed hypersensitivity and other immune responses. Pretreatment biopsies from each patient showed no endothelial expression of endothelial-leukocyte adhesion molecule 1 and only weak to moderate expression of intercellular adhesion molecule 1 and histocompatibility leukocyte antigen-DQ (except for one specimen unreactive with Leu 10). After 5 days of treatment, every patient showed marked endothelial expression of all three antigens (except for the same patient who remained unreactive with Leu 10). Endothelial-leukocyte adhesion molecule-1 expression was confined to postcapillary venular endothelium whereas intercellular adhesion molecule-1 and Leu 10 also were expressed on stromal cells and mononuclear cells. Thus, we conclude that i.v. administration of IL-2 leads to endothelial cell activation. Because IL-2 fails to induce the same antigens on cultured endothelial cells, we infer that IL-2 acts in vivo by inducing the production of other cytokines (e.g., interleukin 1, tumor necrosis factor, lymphotoxin, and interferon-gamma). Finally, since endothelial cell activation at sites of cell-mediated immune responses is well known to result in vascular leakiness to macromolecules, we propose that the vascular leak syndrome accompanying IL-2 therapy may arise from widespread inappropriate endothelial cell activation.