Experimental results and therapeutic strategies. Insulin-dependent diabetes mellitus (IDDM) results from an autoimmune aggression toward beta cells in genetically predisposed individuals. Examination of the frequency of the different antigens coded by the major histocompatibility complex reveals an increased proportion of DR3-DQ2 and DR4-DQ8 haplotypes in IDDM subjects. Sequencing DQ-beta chains in such patients indicates the absence of aspartate in position 57 when compared to control individuals. Islet cell cytoplasmic autoantibodies are early markers of ongoing autoimmunity in addition to insulin autoantibodies before administration of exogenous insulin. Experimental models of autoimmune diabetes like the NOD (NonObese Diabetes) mouse underline the predominant role of T lymphocytes in the constitution of both insulitis and beta cell destruction. In humans, an increased proportion of activated T lymphocytes can be observed but is not specific of the disease. This underlines the need for new cellular markers of the autoimmune process. Transgenic mice allow studies on the consequences of abnormal expression of new molecules on beta cell surface like cytokines or MHC class II molecules which represent a new field of investigation on the pathogenesis of IDDM. Prospective studies in first degree relatives of type I diabetic patients indicate the existence of an asymptomatic phase of beta cell destruction where specific autoimmune markers can be individualized. In some individuals abnormal insulin response to glucose--loss of first phase insulin release during intravenous glucose tolerance test--precedes insulin deficiency. The identification of an autoimmune process leading to beta cell destruction allows new therapeutic approaches with immunointervention at early stages of the disease.