The incidence of destructive pancreatic infiltrates and overt diabetes in animal models of insulin-dependent (type I) diabetes mellitus can be greatly reduced by inactivating or eliminating most T lymphocytes early in life. Because of theoretical and practical concerns about inducing long-term pan-T-lymphocyte inactivation for prevention or treatment of type I diabetes in humans, we hoped that more selective suppression of only the diabetogenic T lymphocyte population might be possible. To this end, two groups suggested that diabetogenic subpopulations of T lymphocytes in NOD mice could be identified by the protein sequence of their T-lymphocyte receptors. This assertion was based on experimental elimination of candidate T-lymphocyte subpopulations in two different short-term models of diabetes induction in NOD mice. For these experiments, identification and elimination of T-lymphocyte subsets were accomplished with monoclonal antibodies that bind specifically to the variable region of the beta-chain (V beta) of the T-lymphocyte antigen receptor and divide the T-lymphocyte pool of the NOD mouse into approximately 20 V beta subsets. To test the relationship between the two T-lymphocyte V beta subsets implicated in these studies and pancreatic beta-cell destruction in unmanipulated animals, both T-lymphocyte subpopulations identified were genetically eliminated from NOD-derived mice by introduction of a mutant T-lymphocyte receptor V beta gene, from which these sequences are genomically deleted. Histological evidence of severe beta-cell destruction and overt diabetes was found in mice homozygous for the deleted V beta gene, indicating that neither V beta gene segment identified in previous studies is required for diabetogenesis in unmanipulated diabetes-prone mice.