Between May 1, 1993 and April 5, 1995, 154 pancreas allograft recipients at 9 institutions were given FK506 posttransplant. Three groups were studied: (1) recipients given FK506 initially for induction and maintenance therapy (n = 82), (2) recipients switched to FK506 for antirejection or rescue therapy (n = 61), and (3) recipients converted to FK506 for other reasons (n = 11). Of 82 patients in the induction group, 7 (9%) had simultaneous bone marrow (BM) and pancreas-kidney (SPK-BM) transplants, 54 (66%) had SPK transplants without BM, 14 (17%) had pancreas transplants alone (PTA), and 7 (9%) had pancreas after previous kidney transplants (PAK). All but 1 recipient was given quadruple immunosuppression (anti-T cell agents plus azathioprine and prednisone) for induction. The median FK506 starting dose was 4 mg/day p.o.; the median average FK506 blood level, 12 ng/ml. The most common side effects were neurotoxicity (16%), nephrotoxicity (13%), and gastrointestinal toxicity (9%). New-onset diabetes mellitus requiring permanent insulin therapy did not occur. Of 61 transplants in the rescue group, 44 (72%) were SPK, 11 (18%) PTA, and 6 (10%) PAK. All but 3 (95%) of the recipients had been on cyclosporine-azathioprine-prednisone triple immunosuppression before substitution of FK506 for cyclosporine; 46% of the recipients had one, and 54% > or = 2, rejection episodes preconversion. The most common side effects were nephrotoxicity (25%), neurotoxicity (23%), and gastrointestinal toxicity (21%). Two recipients were reconverted to cyclosporine because of transient hyperglycemia, and one recipient is on insulin. In the induction group, patient survival at 6 months was 90% for SPK, 100% for PTA, and 100% for PAK. According to a matched-pair analysis, pancreas graft survival for SPK recipients at 6 months was 87% for FK506 versus 70% for cyclosporine recipients (P = 0.04); for PTA recipients, 84% versus 66% (P = n.s.); and for PAK recipients, 80% versus 14% (P = 0.11). When technical failures and death with functioning grafts were censored, pancreas graft survival remained significantly better in the FK506 group. The incidence of first reversible rejection episodes by 6 months in FK506 recipients was 35% for SPK, 40% for PTA, and 20% for PAK. Of 75 pancreas grafts, 64 are currently functioning; in 5 recipients the pancreas failed (1 from rejection); 6 recipients died with a functioning pancreas graft. There were 3 posttransplant lymphomas (all EBV-positive); 2 recipients died and 1 is alive after subtotal colectomy and transplant pancreatectomy. In the antirejection rescue group, patient survival rates at 6 months were 91% for SPK, 100% for PTA, and 80% for PAK (P = n.s.). Pancreas graft survival rates at 6 months were 90% for SPK, 72% for PTA, and 40% for PAK. The incidence of first reversible rejection episodes after conversion to FK506 at 6 months was 44% in SPK, 54% in PTA, and 50% in PAK. Of 61 pancreas grafts, 51 are currently functioning; in 7 recipients the pancreas failed (5 from rejection); 3 recipients died with a functioning graft. There were no posttransplant lymphomas in the rescue group. This multicenter survey shows that FK506 in pancreas transplantation is associated with (1) a low rate of graft loss from rejection when used for induction therapy, (2) a high rate of graft salvage when used for rescue or rejection therapy, and (3) a very low rate of new-onset insulin-dependent diabetes mellitus. These encouraging results are tarnished by 3 posttransplant lymphomas in the induction group; a possible explanation is overimmunosuppression, but further (randomized) studies are necessary to analyze the long-term risk-benefit ratio of FK506 after pancreas transplantation.