Peripheral immunological tolerance is traditionally explained by mechanisms for deletion or inactivation of autoreactive T cell clones. Using an autoimmune disease model combining transgenic mice expressing a well-defined antigen, influenza hemagglutinin (HA), on islet beta cells (Ins-HA), and a T cell receptor transgene (TCR-HNT) specific for a class II-restricted HA peptide, we demonstrate that the conventional assumptions do not apply to this in vivo situation. Double transgenic mice displayed either resistance or susceptibility to spontaneous autoimmune disease, depending on genetic contributions from either of two common inbred mouse strains, BALB/c or B10.D2. Functional studies on autoreactive CD4+ T cells from resistant mice showed that, contrary to expectations, neither clonal anergy, clonal deletion, nor receptor desensitization was induced; rather, there was a non-MHC-encoded predisposition toward differentiation to a nonpathogenic effector (Th2 versus Th1) phenotype. T cells from resistant double transgenic mice showed evidence for prior activation by antigen, suggesting that disease may be actively suppressed by autoreactive Th2 cells. These findings shed light on functional aspects of genetically determined susceptibility to autoimmunity, and should lead to new therapeutic approaches aimed at controlling the differentiation of autoreactive CD4+ effector T cells in vivo.