Type 1 diabetes mellitus is a polygenic disease strongly associated with the class II molecules DR3, 4 and the linked DQ2, 8 alleles. These molecules play an important role in presentation of peptide antigens after intracellular processing to CD4 T lymphocytes. A number of in vitro approaches have been used to elucidate the molecular basis for the association of particular HLA alleles with susceptibility to or protection from Type 1 diabetes mellitus. These have focused on the structure of the antigen-presenting molecules, together with their peptides. Binding studies, peptide elution, molecular modelling and crystallisation of the peptide MHC complex have between them made it possible to define the peptide-binding regions and to examine the stability of binding of peptides from putative autoantigens. It is difficult to study the role of these molecules in vivo in humans, and HLA transgenic mice have been generated to overcome this problem. Studies of mice expressing the HLA class II alleles associated with diabetes have shown that the presence of HLA molecules alone does not cause disease except in the presence of an islet "insult", even when this "insult" would in itself be insufficient to precipitate disease in the absence of the HLA class II transgene. HLA transgenic mice offer a way to elucidate the in vivo role of these molecules, and could help the development of targeted immunotherapy.