MHC class II-binding peptides administered in vivo can be effective and selective inhibitors of T cell activation. To selectively interfere with the activation of autoreactive T cells immunotherapy by peptides can be essentially directed to three cellular targets: APC, autoreactive T cells and regulatory T cells. Experimental models of autoimmune diseases have demonstrated that disease can be prevented or treated by peptides acting on each cellular target. The APC can be functionally inactivated by blocking the binding site of MHC class II molecules, thus preventing antigen presentation to T cells. We have demonstrated that injection of a class II-binding peptide can induce MHC class II blockade in vivo [22], resulting in inhibition of class II-restricted T cell activation and of T cell-dependent antibody responses. This is already a validated approach to prevent experimental autoimmune diseases [3, 32] and it may become applicable to the treatment of human autoimmune diseases. The autoreactive T cells could be inactivated by peptide-specific tolerance induction [14, 24], but it is necessary to define all potential autoantigenic epitopes to ensure complete and effective tolerance. Specific inhibition of autoreactive T cells could also be induced by antigen analogues acting as TCR antagonists [15], but this again requires, at a minimum, detailed knowledge of the autoantigen. Regulatory T cells represent a potentially interesting target for selective immunointervention by peptides. The restricted TCR V gene expression by autoreactive pathogenic T cells in some animal models of autoimmune diseases, for example the shared TCR V gene usage characteristic of EAE [1], raises the possibility of controlling self reactivity at the network level, as originally proposed by Cohen and colleagues [9]. Administration of peptides corresponding to TCR sequences utilized by autoreactive T cells has indeed been reported to down-regulate EAE, presumably via induction of anti-idiotypic T cells with suppressive activity [26, 43]. However, this approach does not always result in suppression of encephalitogenic effector T cells, and it may actually lead to enhanced chronic EAE [16]. It is also possible to induce regulatory T cells with suppressive activity by administration of peptides derived from the candidate autoantigen [24, 39] and this specific peptide therapy has been used to prevent the development of autoimmune diabetes in NOD mice [17]. However, if effective tolerance induction requires knowledge of all autoantigenic epitopes, induction of suppressor T cells by antigen is further complicated by our lack of definite molecular explanations for this phenomenon. Therefore, several immunoregulatory mechanisms induced by administration of class II-binding peptides may be responsible, in vivo, for interference with the activation of antigen-specific T cells. Unfortunately, with the possible exception of MHC blockade, the mechanisms involved are far from being clearly understood. What is needed is a more detailed analysis of these mechanisms in experimental autoimmune models to discern the precise mode of action of immunoregulatory peptides. This knowledge may then be applied to the treatment of human autoimmune diseases, keeping in mind that the clinical situation will invariably be even more complex.