A technique for preparation of pancreatic juice for cytological examination, which was applied in 35 consecutive recipients of a pancreas allograft, is described. Pancreatic juice cytology (PJC) was performed daily, giving a total of 1116 cytological specimens. In these 35 pancreas grafts, 26 acute cellular and two vascular rejections were diagnosed on the basis of exocrine and endocrine graft function, perfusion scans, and clinical course and outcome, and were retrospectively compared with PJC findings. Histology was available only for vascular-type rejections. In uncomplicated cases, mainly neutrophil granulocytes with some ductal epithelial cells and 2-3% lymphocytes were seen. During 96% of acute cellular rejections an increase of lymphocytes was detected to more than 5% of all cells counted. In two-thirds of these episodes, eosinophil granulocytes and/or necrotic epithelial cells were found. The total number of cells was significantly increased in 60% of the cases. A minimum of two of these criteria was necessary for diagnosis of rejection. Rejections were correctly diagnosed in 234 slides (23 episodes) and missed in 36 (three episodes), PJC gave a false-positive result in 27 examinations (three rejections), but a correct negative result in the remaining 819 examinations. This gives a sensitivity of 86.6% and a specificity of 96.8%. With the exception of viral pancreatitis, where no specific changes were seen, PJC also proved very helpful in diagnosing other complications after pancreas transplantation, such as pancreatitis, bacterial or fungal contamination of the juice, and even cyclosporine toxicity. PJC facilitated the detection of isolated rejection episodes of the pancreas graft in recipients of a combined renal pancreatic graft, which happened in six instances. PJC has proved to be a simple but most helpful method for the detection of cellular rejection of pancreas allografts, as well as other complications.