Previous studies have shown that the yield and viability of islets obtained from human or large mammal pancreas depends upon techniques used for islet isolation and upon factors that affect the quality of the donor pancreas. In the present studies, the efficacy of collagenase digestion by ductal perfusion or automated techniques was compared in a canine model of purified islet isolation. The ductal perfusion technique was then developed for human pancreas which was excised before (n = 8) or after (n = 8) multiple organ procurement and without in situ arterial perfusion of hypothermic preservation solution or significant cold storage. Studies with canine pancreas showed that ductal perfusion yielded 2.0 +/- 0.7 microL of islets/g of pancreas and when combined with automated digestion, yields improved to 3.6 +/- 0.8 microL/g (versus perfusion alone, not significant). The yield and viability of human islets was improved when the pancreas was excised before procurement of other donor organs. Results of islet isolation from eight consecutive human pancreases procured in this manner revealed a total yield of 355.2 +/- 44 x 10(3) islets, corresponding to 5345 +/- 600 islets/g (+/- SEM, mean islet diameter 150 microns). Six of eight Ficoll purification attempts succeeded, yielding 186.6 +/- 31 x 10(3) islets of purity ranging from 45-60%. Perifusion with glucose elicited biphasic insulin secretion with a three-fold rise. Islets from two of the isolations were utilized to initiate a clinical trial of islet transplantation in insulin-dependent diabetes mellitus.