Spontaneous insulitis with insulin-dependent diabetes mellitus (IDDM) in rodent models, the BB rat and NOD mouse, has clarified the pathogenesis of and guided decisions on interventional therapy for human IDDM. However, the occurrence in such models of a standard marker of human IDDM, autoantibodies to beta islet cell constituents, has been controversial. Hence we assessed diabetes-prone rodents for the frequencies of raised levels of auto-antibodies to glutamic acid decarboxylase GAD (anti-GAD), insulin and heat shock protein 65 (HSP-65) in relation to levels in non-diabetes-prone animals and levels in human diabetic sera. Assays were performed sequentially at various ages of life. The immunoassays used for anti-GAD and anti-insulin were those validated for sensitivity and specificity for detection of the corresponding autoantibodies in human IDDM sera at international workshops. Positive controls included human IDDM sera with reactivity with GAD or insulin and, for mouse anti-GAD, the highly reactive monoclonal antibody, GAD-6. The results were that levels of autoantibodies in diabetes-prone BB rats or NOD mice to the "IDDM-relevant' autoantigens in our panel did not exceed levels in control rats or mice, and were much lower than levels in humans with IDDM. We conclude that the BB rat and NOD mouse represent a model, but not a facsimile, of human IDDM and that therapeutic successes in such models should be interpreted with caution in relation to interventional therapy for human IDDM.