The pathogenesis of diabetes mellitus (DM) type I in dogs is still largely unknown but it is suspected, that the immune system plays an important role in the development of DM. The aim was to develop a specific and reliable method for the estimation of AA against insulin in dogs. A newly develooped ELISA (streptavidin coated microtitre plates with porcine insulin coupled to biotin) was compared with immunohistochemistry (IHC) carried out on canine and porcine pancreatic tissue. Serum samples of diabetic dogs (n=43) were allotted to group A (prior to insulin therapy; n=21), group B (after the onset of therapy with porcine insulin lente; repeated samples taken from several dogs, n=34) and healthy dogs (n=101) and examined for AA. With the ELISA, 3 dogs of group A, and 6 dogs of group B were found positive for insulin AA. IHC showed 1 positive result in group A, and 10 positive results in group B. However, after blocking with insulin in 6 dogs there was still a positive reaction with single cells in IHC whereas in the ELISA all positive reactions could be blocked. In the group of healthy dogs 99 had no insulin AA. The results of the ELISA matched well with those of IHC (94 %). With IHC insulin was not the only antigen bound by serum AA of diabetic dogs. Further studies are required to estimate if these AA are present before the onset of clinically evident diabetes.