CD4+ T cells recognize antigenic peptides bound to the polymorphic peptide-binding site of major histocompatibility complex (MHC) class II molecules. The polymorphism of this site is thought to dictate which peptides can be bound and thus presented to the T cell receptor. The mycobacterial 65-kDa heat-shock protein (hsp65) peptide 3-13 is an important T cell epitope: it is immunodominant in the mycobacterium-specific T cell response of HLA-DR3+ individuals but, interestingly cannot be recognized in the context of any other HLA-DR molecules. We, therefore, have tested whether the hsp65 epitope p3-13 is selected for T cell recognition in the context of only HLA-DR3 molecules by an unique binding specificity for HLA-DR3. Using biotinylated peptides and EBV-transformed BLCL comprising all known HLA class II specificities, we find that p3-13 binds to HLA-DRw17(DR3) but not to any other HLA-DR molecule. Conversely, a control peptide p307-319 influenza hemagglutinin binds to all known HLA-DR molecules but only weakly to HLA-DRw17 and HLA-DR9. Peptide binding could be inhibited by excess unbiotinylated competitor analogue as well as by anti-DR monoclonal antibodies but not by anti-class I-, anti-DP- or anti-DQ monoclonal antibodies. The amino acid sequence of DRw17 molecules differs uniquely at five positions from the other DR beta 1 sequences. Three of these five residues (positions 26, 71 and 74) are potential peptide contacting residues. These residues map closely together in the hypothetical three-dimensional model of the DR molecule and, thus, most probably form a positively charged pocket, critical for the binding of p3-13. Interestingly, p3-13 does not bind to a DR3 variant, the DRw18 molecule. The DRw18 beta 1 chain differs from DRw17 at two major positions, close to or within the DRw17-specific pocket. These substitutions drastically change the structure and charge of the pocket and thus presumably abrogate its ability to bind p3-13.