BACKGROUND We compared the ability of impure, dispersed pancreatic islet tissue (DPIT) and purified islets to engraft in the peritoneal cavity of dogs. We also tested whether posttransplantation insulin therapy affects islet engraftment. METHODS Thirty-two dogs underwent total pancreatectomy. DPIT was autotransplanted intraperitoneally in nine dogs. Purified islets were autotransplanted intraperitoneally in 13 dogs and intraportally in seven dogs. Dogs received comparable islet mass. One half of the recipients of intraperitoneal grafts received 12 days of posttransplantation exogenous insulin. Three dogs did not undergo transplantation. Intravenous glucose tolerance tests were done at 60 and 180 days. RESULTS The long-term graft functional survival rate of intraperitoneal DPIT graft was significantly better than the rate of intraperitoneal purified islets (p < 0.01) and was as good as the rate of intraportal purified islets. Purified islets transplanted intraperitoneally failed early compared with other groups, with only 46% functioning at 3 weeks (p = 0.05); exogenous insulin reduced this early failure rate (p = 0.05). At 6 months 67% of intraperitoneal DPIT and 86% of intraportal purified grafts were functioning. Glucose disposal did not differ between groups. CONCLUSIONS The peritoneal cavity is a safe, practical site for islet transplantation, particularly for high-volume DPIT grafts. DPIT may provide a larger, more viable beta-cell mass than purified islets, or the acinar-ductal tissue may have a positive effect on engraftment. Exogenous insulin may promote engraftment of transplanted islets with a marginal beta-cell mass.