Currently it is not feasible to isolate sufficient numbers of islets from a single pancreas for clinical transplantation. We examined whether small numbers of islets obtained from multiple donors could be used for transplantation. Islets were isolated from four inbred strains of mice (DBA/2, DBA/1, C3H, and A.SW) by a stationary collagenase digestion and Ficoll separation and transplanted into the renal subcapsular space of streptozotocin-induced diabetic B6AF1 mice. At least 200 handpicked islets were required to produce normoglycemia in syngeneic and allogeneic diabetic recipient mice. None of the mice given 50 islets became normoglycemic within 2 weeks postgrafting. When various numbers of purified islets from a single donor were transplanted, the survival was significantly better for 200-islet allografts than for 800-islet and 400-islet allografts. When a 200-islet composite graft was prepared from four donors (50 fresh handpicked islets from each donor), the survival of the composite graft as measured by sustained normoglycemia in nonimmunosuppressed recipients was dramatic, with 17 of 18 recipients maintaining normoglycemia indefinitely (greater than 200 days). Similarly, when islets isolated from four donors and cultured for various periods were mixed and transplanted (200 islets/recipient) all recipient mice (n = 8) enjoyed indefinite graft survival. Use of higher numbers of purified islets or crude islets in a composite multiple-donor islet allograft was less effective in achieving indefinite graft survival. Thus, transplantation of a composite graft made up with subtherapeutic numbers of islets from multiple histoincompatible donors to provide adequate therapeutic numbers is a practical solution to the lack of islet availability. In addition, composite islet grafts appear to possess immunological advantages, with significantly prolonged survival over that produced by single-donor islet grafts.