After pancreas-kidney transplantation, it is difficult to obtain an accurate estimate of the insulin secretion of the pancreas graft, since several pitfalls are involved using peripheral C-peptide and/or insulin measurements in this determination. In this study, the individual kinetic parameters of C-peptide and then the rates of insulin secretion were estimated by two mathematical methods, the deconvolution method and the "combined model" during slow (oral glucose) and fast (intravenous glucagon) changes in insulin secretion in six successful pancreas-kidney transplant recipients with systemic delivery of insulin (Px), six nondiabetic kidney-transplant recipients with portal insulin secretion (Kx), six nondiabetic controls (NS), and six C-peptide-negative insulin-dependent diabetes mellitus patients (IDDM). Decreased C-peptide clearance and basal and poststimulatory hyperinsulinemia were found in both Px and Kx compared with NS (P < 0.05). Similar glucose responses were observed after intravenous glucagon in all groups, whereas the responses after oral glucose were 30% higher in Px and Kx than in NS (P < 0.05). During oral glucose and after intravenous glucagon, both mathematical methods resulted in significantly lower maximal and incremental insulin secretion rates (ISR) in Px than in Kx (P < 0.05). In contrast, calculations of incremental ISR in NS and Px induced by the two -cell stimuli were about the same but significantly higher in Kx than in NS (P < 0.05). These results differed markedly from those obtained using peripheral measurements of insulin and C-peptide alone. In conclusion, when C-peptide clearance and insulin metabolism change, such as in pancreas-kidney transplant recipients, accurate evaluation of insulin secretion from the graft can be obtained only by using individual kinetics of the peptides before calculating the ISR. This study also clearly demonstrates that insulin secretion after pancreas transplantation is still defective.