A major dilemma in pancreas transplantation is the lack of reliable methods for the early detection of allograft rejection. Over a 26-mo period, 70 rejection episodes occurred in 36 patients (13 isolated-pancreas, 23 simultaneous pancreas-kidney recipients) with pancreaticoduodenocystostomy. A total of 300 radionuclide pancreas examinations were performed (mean 8.3/patient) utilizing 99mTc-labeled DTPA. Computer analysis generated a quantitative measure of blood flow to the allograft (technetium index, TI). Rejection episodes were defined as isolated pancreas, isolated kidney, or combined pancreas-kidney. Mean urinary amylase (UA) levels and TI during normal allograft function were 30,256 U/L and 0.57%, respectively, whereas levels heralding rejection were 6873 U/L and 0.39% (P less than .05). The treatment of rejection based on kidney dysfunction or combined pancreas-kidney dysfunction resulted in significantly higher graft salvage and a lower incidence of hyperglycemia compared with isolated-pancreas-allograft rejection. After therapy, a TI greater than 0.3% was associated with 95.9% graft survival, whereas levels less than 0.3% resulted in a 72.7% rate of graft loss (P less than .001). Similarly, pancreas allografts with a UA greater than 10,000 U/L had 92.2% functional survival, whereas levels less than 10,000 U/L resulted in a 53.3% rate of graft loss (P less than 0.001). Overall, reversal of rejection occurred in 80% of cases, with 10 pancreas and 2 kidney allografts lost due to rejection. Monitoring pancreas-allograft function by UA, TI, and renal function in simultaneous transplants allows for the timely diagnosis and successful treatment of pancreas-allograft rejection.