Our studies to date indicate that treatment with anti-idiotype to monoclonal anti-MHC antibodies can markedly influence the repertoire of anti-MHC antibodies expressed. The antibodies discussed here appear to represent two classes, one of which is public, probably representing expression of a germ line gene, and the second of which probably represents either a somatic variant of a germ line gene or one of a very large number of germ line genes devoted to the same specificity. In either case, this class of idiotype arises only rarely following antigen, but is readily induced by anti-idiotype treatment. There may indeed exist a third class of anti-MHC monoclonal antibodies representing distant somatic diversification from a germ line gene. Our only indication of this so far is that certain idiotypes are only induced after multiple boosts with anti-idiotypes rather than a single treatment. This finding, however, may reflect again the enormous number of different ways in which anti-MHC antibodies to the same nominal specificity can be produced. Finally, our results in adoptive transfer systems indicate that manipulation of idiotype expression by anti-idiotype treatment probably involves a complex pathway of cellular interactions. If, as we expect, these intercellular interactions involve idiotype and/or anti-idiotypic receptors, they should provide a model for mechanistic studies of the in vivo immune network.