The major histocompatibility complex class II genes play an important role in the genetic predisposition to many autoimmune diseases. In the case of rheumatoid arthritis (RA), the human leukocyte antigen (HLA)-DRB1 locus has been implicated in the disease predisposition. The "shared epitope" hypothesis predicts that similar motifs within the third hypervariable (HV3) regions of some HLA-DRB1 alleles are responsible for the class II-associated predisposition to RA. Using a line of transgenic mice expressing the DQB1*0302/DQA1*0301 (DQ8) genes in the absence of endogenous mouse class II molecules, we have analyzed the antigenicity of peptides covering the HV3 regions of RA-associated and nonassociated DRB1 molecules. Our results show that a correlation exists between proliferative response to peptides derived from the HV3 regions of DRB1 chains and nonassociation of the corresponding alleles with RA predisposition. While HV3 peptides derived from nonassociated DRB1 molecules are highly immunogenic in DQ8 transgenic mice, all the HV3 peptides derived from RA-associated DRB1 alleles fail to induce a DQ8-restricted T-cell response. These data suggest that the role of the "shared epitope" in RA predisposition may be through the shaping of the T-cell repertoire.