This follow-up multicenter analysis is based on 362 pancreas allograft recipients at 14 institutions who were given tacrolimus between 1 May 1994 and 15 November 1995. Three groups were studied: (1) recipients given tacrolimus initially for induction and maintenance therapy (n = 250; 215 without, 35 with, a concurrent bone marrow transplant), (2) recipients who converted to tacrolimus for rescue or rejection therapy (n = 89), and (3) recipients who converted to tacrolimus for other reasons (n = 23). Of 215 recipients without a bone marrow transplant in the induction group, 166 (77%) underwent a simultaneous pancreas-kidney transplant (SPK), 29 (14%) a pancreas transplant alone (PTA), and 20 (9%) a pancreas after previous kidney transplant (PAK). Initial antibody therapy was given to 185 (86%) recipients. All 215 received tacrolimus and prednisone; 202 (94%) also received azathioprine (AZA) and 11 (5%) mycophenolate mofetil (MMF). The most common side effects of tacrolimus were neurotoxicity in 21%, nephrotoxicity in 21%, gastrointestinal (GI) toxicity in 13%, and diabetogenicity in 13% of these recipients. No recipient in this group developed new-onset insulin-dependent diabetes mellitus. Of 89 recipients in the rescue group, 71 (79%) had an SPK, 11 (13%) a PTA, and 7 (8%) a PAK. Before conversion, all had been on cyclosporine (CsA)-based immunosuppression; 74% of them had 2 or more rejection episodes previously. The most common side effects were nephrotoxicity in 27%, neurotoxicity in 26%, GI toxicity in 18%, and diabetogenicity in 8% of these recipients. No recipient in this group developed new-onset insulin-dependent diabetes mellitus. In the induction group patient survival at 1 yr was 98% for SPK, 79% for PTA, and 100% for PAK recipients. According to a matched-pair analysis, pancreas graft survival for SPK recipients at 1 yr was 88% with tacrolimus vs. 73% with CsA (p = 0.002); for PTA recipients, 68% vs. 70% (p > 0.35); and for PAK recipients, 85% vs. 65% (p = 0.13). Graft loss from rejection was not different with tacrolimus vs. CsA in all 3 pancreas recipient categories. At 1 yr, 17% of recipients had converted from tacrolimus to CsA for diabetogenicity, nephrotoxicity, or rejection; 23% had converted from AZA to MMF. The incidence of post-transplant lymphoma was < 2%. In the rescue group, patient survival rates at 1 yr were 96% for SPK, 100% for PTA, and 86% for PAK recipients (p < 0.08). Pancreas graft survival at 1 yr was 89% for SPK, 58% for PTA, and 69% for PAK recipients (p = 0.004). Graft loss from rejection was significantly lower for SPK vs. PTA or PAK recipients. At 1 yr, 20% of recipients had reconverted from tacrolimus to CsA for rejection, neurotoxicity, or nephrotoxicity; 19% had converted from AZA to MMF. There were no post-transplant lymphomas in the rescue group. This follow-up multicenter analysis shows that tacrolimus after pancreas transplantation is associated with high graft survival rates when used for induction and with high graft salvage rates when used for rescue therapy. The rate of graft loss from rejection is low in all 3 pancreas recipient categories. The overall incidence of new-onset insulin-dependent diabetes mellitus is < 1%, as is the incidence of post-transplant lymphoma. Converting from tacrolimus to CsA and, in patients on tacrolimus, from AZA to MMF, is safe; interchangeable use of drugs appears to be of immunologic benefit. To determine the best immunosuppressive regimen after pancreas transplantation, a prospective randomized study comparing tacrolimus and MMF vs. Neoral plus MMF is mandatory.