Antigen-specific T cell activation depends initially on the interaction of the T cell receptor (TCR) with peptide/MHC. In addition, a costimulatory signal, mediated by distinct cell surface accessory molecules, is required for complete T cell activation leading to lym-phokine production and proliferation. CD28 has been implicated as the major receptor on T cells responsible for delivering the costimulatory signal. Although two distinct ligands for CD28, B7-1 and B7-2, have been identified on antigen-presenting cells (APC), the costimulatory role of each molecule during a physiological immune response remains unresolved. In the present study, the relative roles of B7-1 and B7-2 interactions were evaluated in an allogeneic pancreatic islet transplant setting. In isolation, anti-B7-2 mAbs and, to a much lesser degree, anti-B7-l mAbs suppressed T cell proliferative responses to allogeneic islets or splenic APC in vitro. Maximal inhibition of the allogeneic response was observed using a combination of the anti-B7-l and anti-B7-2 mAbs. Administration of anti-B7-2 but not anti-B7-l mAbs prolonged C3H allograft survival in B6 recipients, with a combination of both mAbs significantly prolonging rejection beyond either mAb alone. The immunosuppressive effects of the in vivo mAb treatment were not manifested in in vitro analyses as T cells isolated from suppressed mice responded normally to allogeneic stimuli in terms of both proliferation and lymphokine production. However, combined mAb therapy in vivo selectively delayed CD4+ T lymphocyte infiltration into the graft. These data suggest that both B7-1 and B7-2 costimulatory molecules are active in vivo, although B7-2 plays a clearly dominant role in this allograft model. The mechanism of immune suppression in vivo remains unresolved but may occur at sites distinct from the allograft.