One explanation offered for the uniquely high precursor frequencies of T cells which recognize allogeneic major histocompatibility complex (MHC) molecules, and their lack of self-MHC restriction, is that the alloreactive cells are polyclonal populations the primary specificity of which is self-MHC plus peptide X1, X2, ... Xn. These are postulated to cross-react with allo-MHC plus peptides Y1, Y2, ... Yn. It has been further suggested that the structural basis for the crossreactivity between different MHC alleles is the similarity in amino acid sequence of that part of the molecule predicted to make contact with the T cell receptor (TcR). In order to test this concept, T cells were obtained with dual specificity for influenza haemagglutinin (HA), restricted by HLA-DR1Dw1, and for DR4Dw4/Dw14 expressed on allogeneic human B cell lines, and the specificity of one clone was studied in detail. The exposed, TcR-contacting surfaces of these two DR molecules are predicted to be identical. Although the HA-specific response was stimulated by DR1-expressing mouse DAP.3 transfectants, DAP.3 cells expressing the alloantigen DR4Dw4 were unable to stimulate, possibly because of a failure to present the necessary human peptide for anti-DR4 allorecognition. Therefore, the effects of pulsing the DR4Dw4-expressing DAP.3 cells with the HA peptide were examined. This peptide is known to bind to both DR1 and DR4. Addition of the HA peptide restored the anti-DR4Dw4 response. These data support the concept that allorecognition in some responder/stimulator combinations can be explained by cross-reactivity at the level of the MHC molecule and the peptide.