Determination of the long-term function of islet transplantation in relation to the implantation site and the numbers of islets is of scientific interest and, with human islet transplant trials in progress, is a pressing clinical question. In this study, highly purified canine islets were isolated by collagenase digestion and Ficoll purification, and autotransplanted into either the spleen (in 10 dogs) or the liver (in 12 dogs). Dogs transplanted with islets into the spleen or liver received 264,300 +/- 20,300 (mean +/- SEM) and 158,600 +/- 15,000 islet equivalents (150-microns-sized islets) respectively. Graft survival at 1 yr was 86% in intrasplenic islet autografts (ISTx) and 50% in intraportal islet autografts (IPTx). Intravenous glucose tolerance tests and mixed meal-oral glucose tests were performed 1-12 mo from islet transplantation. Compared to controls, ISTx and IPTx dogs showed a similar decrease of glucose tolerance after both intravenous glucose tolerance tests and mixed meal-oral glucose tests. On intravenous glucose tolerance tests, plasma insulin levels were lower in ISTx than in IPTx dogs and controls. On mixed meal-oral glucose tests, insulin values were higher in IPTx dogs than in controls. There was a positive correlation (r = .56, p < 0.05) between the number of transplanted islet equivalents and the K values. These results demonstrate that, in dogs with islet transplant: 1) long-term islet survival can be achieved in the spleen better than in the liver; 2) islet survival is related to the mass of transplanted islets in the spleen, but not in the liver, where other factors probably affect islet survival; 3) the ability of metabolizing glucose is reduced after both intrasplenic and intraportal islet autografts; 4) both reduced insulin secretion (predominant in ISTx dogs on intravenous glucose tolerance testing) and insulin resistance (predominant in IPTx dogs on mixed meal-oral glucose tests) are the probable causes of the decreased glucose tolerance.