Although we and others have generated IRS-2 knock-out (IRS-2(-/-)) mice, significant differences were seen between the two lines of IRS-2(-/-) mice in the severity of diabetes and alterations of beta-cell mass. It has been reported that although IRS-1 and IRS-3 knock-out mice showed normal blood glucose levels, IRS-1/IRS-3 double knock-out mice exhibited marked hyperglycemia. Thus, IRS-1 and IRS-3 compensate each other's functions in maintaining glucose homeostasis. To assess the effect of genetic background and also ablation of IRS-3 on IRS-2(-/-), we generated IRS-2/IRS-3 double knock-out (IRS-2(-/-)IRS-3(-/-)) mice by crossing IRS-3(-/-) mice (129/Sv and C57Bl/6 background) with our IRS-2(-/-) mice (CBA and C57Bl/6 background). Intercrosses of IRS-2(+/-)IRS-3(+/-) mice yielded nine genotypes, and all of them including IRS-2(-/-)IRS-3(-/-) mice were apparently healthy and showed normal growth. However, at 10-20 weeks of age, 20-30% mice carrying a null mutation for the IRS-2 gene, irrespective of the IRS-3 genotype, developed diabetes. When mice with diabetes were excluded from the analysis of glucose and insulin tolerance test, IRS-2(-/-)IRS-3(-/-) showed a degree of glucose intolerance and insulin resistance similar to those of IRS-2(-/-) mice. Both IRS-2(-/-) and IRS-2(-/-)IRS-3(-/-) mice had moderately reduced beta-cell mass despite having insulin resistance. Insulin-positive beta-cells were decreased to nearly zero in IRS-2(-/-) mice with diabetes. Although Pdx1 and glucose transporter 2 expressions were essentially unaltered in islets from IRS-2(-/-) mice without diabetes, they were dramatically decreased in IRS-2(-/-) mice with diabetes. Taken together, these observations indicate that IRS-3 does not play a role compensating for the loss of IRS-2 in maintaining glucose homeostasis and that the severity of diabetes in IRS-2(-/-) mice depends upon genetic background, suggesting the existence of modifier gene(s) for diabetes in mice of the 129/Sv genetic strain.