Genetic analysis of the development of diabetes and insulitis has been performed in the nonobese diabetic (NOD) mouse strain, a model of insulin-dependent (type I) diabetes mellitus. (NOD X C57BL/10)F1, F2, and (F1 X NOD) first-, second-, and third-backcross generations were studied. The data obtained were consistent with the hypothesis that diabetes is controlled by at least three functionally recessive diabetogenic genes, or gene complexes, one of which is linked to the MHC of the NOD. In contrast, pancreatic inflammation leading to insulitis was found to be controlled by a single incompletely dominant gene. One of the two diabetogenic loci that is not linked to the MHC appears to be essential for the development of severe insulitis. This diabetogenic gene may be identical to the gene that controls the initiation of the autoimmune response that progresses to insulitis. Although this gene appears to be functionally recessive in its control of diabetes, it is incompletely dominant in its control of insulitis. The MHC-linked diabetogenic gene, although not required for the development of insulitis, apparently influences the progression of the autoimmune response since NOD MHC homozygotes in the backcross generations displayed the highest incidence and most severe cases of insulitis. Interestingly, we have found two MHC heterozygous backcross females that have become diabetic, suggesting that either the MHC-linked diabetogenic gene is not strictly recessive or that a recombination event has occurred between the diabetogenic gene and the K or I-A regions of the MHC. The third diabetogenic locus appears to influence the progression of severe insulitis to overt diabetes. In animals homozygous at this locus, diabetes may result from a decreased ability to develop a protective suppressor response to the autoimmune process.