As organ-specific autoimmune diseases do not become manifest until well-advanced, interventive therapies must inhibit late-stage disease processes. Using a panel of immunogenic peptides from various beta cell Ags, we evaluated the factors influencing the efficacy of Ag-based therapies in diabetes-prone NOD mice with advanced disease. The ability of the major beta cell autoantigen target determinants (TDs) to prime Th2 responses declined sharply between 6 and 12 wk of age, whereas the ability of immunogenic ignored determinants (IDs) of beta cell Ags to prime Th2 responses was unaffected by the disease process. The different patterns of TD and ID immunogenicity (even from the same beta cell Ag) may be due to the exhaustion of uncommitted TD-reactive, but not ID-reactive, T cell pools by recruitment into the autoimmune cascade. Therapeutic efficacy was associated with a peptide's immunogenicity and ability to promote Th2 spreading late in the disease process but not its affinity for I-Ag7 or its expression pattern (beta cell specific/nonspecific or rare/abundant). Characterization of some IDs revealed them to be "absolute" cryptic determinants. Such determinants have little impact on T cell selection, leaving large precursor T cell pools available for priming by synthetic peptides. Traditional Ag-based therapeutics using whole autoantigens or their TDs cannot prime responses to such determinants. These findings suggest a new strategy for designing more efficacious Ag-based therapeutics for late-stage autoimmune diseases.