In the response of T cells to foreign antigens, the ligand for the T cell alpha/beta receptor is presented on a cell surface as a fragment of antigen complexed to one of the membrane molecules encoded in the major histocompatibility complex (MHC). The receptor apparently interacts via its variable elements (V beta, D beta, J beta, V alpha and J alpha) with residues within both the antigen and MHC portion of the ligand. The frequency of T cells responding to a conventional antigen plus self MHC is usually quite low, presumably reflecting the relative rarity of receptors with the particular combination of variable elements to match the antigen/MHC ligand. T cells also respond to allogeneic forms of MHC molecules in the absence of added antigen. In this case the frequency of responding T cells is very high. One hypothesis to explain this observation is that, in the absence of foreign antigen, MHC molecules are complexed to a large array of peptides derived from self-proteins. In this case the combination of the polymorphic MHC amino acid residues and many different self peptides presents so many possible ligands that the likelihood of recognition by a given T cell receptor is quite high. The recent crystallography experiments which revealed a dramatic binding cleft on the face of a human MHC molecule have given impetus to this view, but as yet there is no direct supporting evidence. We have recently described a close association between murine T cell receptors utilizing the V beta 17a element and reactivity to various allogeneic forms of the murine MHC molecule, I-E (ref. 8). In this paper, we show that this I-E ligand is detected on B cells, but not on I-E+ macrophages or fibroblasts expressing a transfected I-E gene. These results strongly suggest a B cell specific product combines with I-E to form the allogeneic ligand for V beta 17a+ receptors and thus support the concept of alloreactivity described above.