On the basis of our extensive studies on collagen induced arthritis in HLA class II transgenic mice, we proposed a hypothesis to explain role of shared epitope in rheumatoid arthritis (RA) association. According to our hypothesis, complementation between both DQ and DR molecules is required for susceptibility or protection from disease. While certain DQ alleles predispose individuals to RA, DRB1 molecule can modulate disease by shaping T-cell repertoire in the thymus by providing self-peptides and presented by DQ molecules. Using A beta o.DQ8 transgenic mice, we tested ability of peptides derived from HV3 of DR molecules, implicated in RA positively or negatively, to activate T cells. While the peptides derived from RA susceptible DR molecule were poor binders and poor in activating T cells, the peptides derived from RA resistant DR molecules were high affinity binders and efficient T-cell activators. Our experiments suggest that high affinity DR peptides could induce tolerance to autoimmunity while the low affinity peptides could be permissive to autoimmunity. Using peptide from DRB1*0402 molecule, known to be associated with resistance to RA, prior to induction of collagen induced arthritis prevents the onset of disease. Thus, self-peptides derived from HLA molecules could potentially generate tolerance or autoimmunity depending on their binding affinity with HLA molecules.