In the United States diabetes in now the principle cause of end-stage renal disease. For diabetic patients undergoing cadaveric kidney transplantation, a combined kidney-pancreas (KP) transplant is often recommended because this option is perceived to carry no additional risk. However, most transplant centres have restricted KP transplantation to patients with few diabetic complications and no coronary artery disease. We compared survival rates after KP transplantation with those after kidney transplantation alone in clinically similar though non-randomised patient groups. In 173 consecutive diabetic renal transplant candidates, 3-year patient survival in 54 KP recipients was 68%, versus 90% in 46 patients who received a cadaveric kidney alone (p = 0.01). The remaining patients had a living-related-donor kidney transplant, either alone (65) or followed 4-20 months later by a pancreas transplant (8), with survival similar to that with a cadaveric kidney. Independent variables associated with early death were age, history of congestive heart failure, and pancreas transplantation. A serious complication of pancreas transplantation was infection, or which 14 of 54 recipients required pancreatectomy; KP recipients had a higher death rate from infection in the first 12 months (p = 0.034). In view of the excess mortality associated with KP transplantation, we suggest that the combined operation should be reserved for young patients with no history of congestive heart failure, or for patients in whom hyperglycaemia is life-threatening. A randomised trial is needed to compare the long-term outcomes of these procedures.