Pancreatic distension with collagenase solution followed by stationary in vitro digestion yields large numbers of intact islets. We compared in rats two routes of collagenase injection, pancreatic ductal (PD) and portal venous (PV), for islet yield, in vitro insulin secretory capacities, and in vivo functional viability. The islet yield in the PD method (n = 11) was greater than that in the PV method (n = 8) (682 +/- 27 vs. 417 +/- 39 per pancreas, P less than 0.025). The insulin release from the PD islets in response to 16.7 mM glucose increased gradually following culture, 3.2 +/- 0.8 ng/10 islets/30 min (fresh) to 12.3 +/- 2.1 (24-hr culture). In contrast, insulin release from the PV islets increased during the first 6 hr of culture, but decreased after 24 hr in culture. Under electronmicroscopic examination, the PD islets revealed a well preserved structure with healthy endocrine cells, while the PV islets showed a dilated capillary network and distorted endocrine cell continuity. When 100 PD islets were transplanted into streptozotocin-induced diabetic B6AF1 mice (n = 8), all the recipient mice restored normoglycemia (less than 200 mg/dl) within 1-4 days following transplantation and maintained it until rejection. However, the recipient mice given 100 PV islets showed a significant delay in restoring normoglycemia, and 3 of 8 mice given 100 PV islets were still hyperglycemic on day 4 postgrafting. In summary, pancreatic ductal collagenase injection followed by stationary in vitro digestion reproducibly yields higher numbers of intact and viable islets when compared with portal venous collagenase injection, indicating the superiority of this method to portal venous injection.