Forty-four clinical episodes of suspected (pancreas) transplant rejection in 17 pancreatic transplantation patients were reviewed retrospectively. The clinical impression of acute graft rejection, chronic rejection, or nonrejection in each episode was correlated with the results of 19 nuclear medicine, 12 ultrasound (US), and 44 magnetic resonance (MR) imaging studies. US was found to be a moderately sensitive (82%) method of detecting graft rejection. US also was effective in identifying intra- and peripancreatic fluid accumulations. Nuclear medicine imaging was also a sensitive technique (86%) and the only modality that provided physiologic information regarding graft perfusion. MR imaging allowed correct prediction of the presence or absence of graft rejection in 39 of 44 cases (sensitivity, 100%; specificity, 76%) and was an effective means of detecting pathologic fluid collections. Nuclear medicine, US, and MR imaging are all believed to be sensitive methods of detecting graft rejection and are complementary adjuncts to the clinical evaluation of pancreatic transplants.