A mathematical nonlinear regression model of several parameters (baseline insulin intake, posttransplant 2-h postprandial blood glucose, and stimulated C-peptide) from type 1 diabetics with HbA1c <6.5% who do not require insulin therapy and have no hypoglycemic instances was developed for accurately predicting supplemental insulin requirements in the posttransplant period. An insulin deficit threshold of 0.018 U/kg/day was defined as the average first-year calculated insulin deficit (CID), above which HbA1c rose to >6.5% during year 2 of the posttransplant period. When insulin-untreated subjects were divided into two groups based on whether the average CID was smaller (group I) or greater (group II) than the insulin deficit threshold, HbA1c was found to be similar in the two groups in year 1, but increased significantly in group II to above 6.5% (with mean glucose of 121.9 mg/dl) but remained below 6.5% in group I subjects (with mean glucose of 108.7 mg/dl) in year 2 of the follow-up period. The greater insulin deficit in group II was also associated with a higher susceptibility to hyperglycemia during periods of low serum Rapamune and Prograf levels (combined levels below 11.2 and 4.7 ng/ml, respectively). Although the differences between predicted insulin requirement (PIR) and actual empirical insulin intake in the insulin-treated subjects were generally small, they were nonetheless sufficient to identify over- and underinsulinization at each follow-up visit for all subjects (n = 14 subjects, 135 observations). The newly developed model can effectively identify underinsulinized islet transplant recipients at risk for graft dysfunction due to inadequate supplemental insulin intake or those potentially susceptible to graft function loss due to inadequate immunosuppression. While less common following islet cell therapy, the model can also identify overinsulinized subjects who may be at risk for hypoglycemia.