The aims of this study were to devise an omental pouch site for islet implantation in a preclinical large animal and to compare the function of islets engrafted to this site with islets implanted into the spleen. Highly purified islets were isolated from outbred mongrel dogs, then grafted into totally pancreatectomized outbred recipients. Autografts of islets were implanted into a greater omental pouch (group [gp] 1, n = 12) or into the spleen by venous reflux (gp 2, n = 12). Allografts of single donor islets were implanted into the omental pouch (gp 3) of dogs that received CsA (n = 9) or untreated controls (n = 3). The threshold islet mass that consistently reversed diabetes in gp 1 was 10 microliters/kg, which exceeded by 2.5-fold that required in gp 2. Normoglycemia was induced and maintained for 2 months in 6 gp 1 and 8 gp 2 dogs. At IVGTT, the K value (decline in glucose, %/min) was 1.3 +/- 0.4 in gp 1 versus 1.5 +/- 0.2 in gp 2 (P > 0.2). Peripheral venous insulin levels were significantly lower (P < 0.01) in gp 1. Omentectomy (gp 1) or splenectomy (gp 2) induced prompt hyperglycemia. All gp 3 dogs (received > 10 microliters/kg) were initially normoglycemic: grafts of untreated controls failed at 4.2 +/- 1.8 days. In 1 CsA-treated dog the graft failed for technical reasons; normoglycemia persisted in the other 8 for 10, 15, and 21 days, and in 5 instances for > 30 days. When CsA therapy was stopped at 30 days, normoglycemia persisted for 34 +/- 9.5 days. We conclude that purified islets restore normoglycemia after implantation into the omental pouch of diabetic dogs. Compared with intrasplenic islet implantation, an increased graft volume is required and insulin levels are lower.