Immunosuppression with B7 antagonists might have 2 opposite effects: reducing T-cell costimulation through CD28 but also preventing CTLA-4 from transmitting its negative regulatory signal. We therefore hypothesized that a selective blockade of CD28 might be qualitatively different from blocking B7. It was previously reported that CD28 modulation prolongs allograft survival in the rat and reverses induction of experimental autoimmune encephalomyelitis in mice. However, whether CD28 or B7 blockade results in similar immunosuppression on alloimmune and self-restricted responses to soluble antigens has not yet been investigated. Here, we addressed this issue in vitro with antagonist anti-CD28 Fab fragments and in vivo using the modulating anti-rat JJ319 monoclonal antibody. As in the inhibition of B7 with CTLA4 immunoglobulin, anti-CD28 Fab fragments inhibited allogenic T-cell proliferation in mixed cultures. In vivo modulation of CD28 blocked the expansion of alloreactive T cells and promoted their apoptosis. In contrast, selective blockade of CD28 did not modify T-cell proliferative responses and antibody production to soluble antigens, whereas blocking B7 with CTLA4 immunoglobulin did. Our data show that blocking CD28, while leaving CTLA4-B7 interactions undisturbed, inhibits alloreactive CD4+ T-cell expansion but does not modify the response to nominal antigens presented in the context of a self-major histocompatibility complex. That B7 engagement is needed for self-restricted responses whereas engagement of CD28 is not essential adds to the suggestion that another unidentified ligand of B7 might deliver a costimulatory signal in the absence of CD28.