In light of recent technologic advances, we developed a canine renal allograft model utilizing implantable, programmable infusion pumps and biocompatible catheters to reexplore the concept of local immunosuppres-sion. Thirteen mongrel dogs underwent bilateral nephrectomy and autotransplantation of 1 kidney via end-to-end renal-iliac artery and end-to-side renal-iliac vein anastomoses. The proximal end of an infusion catheter directed into the iliac artery was tunneled to a subcutaneously placed programmable pump. A second, sampling catheter was placed with its tip in the iliac vein just proximal to the venous anastomosis. During a period of i.a. infusion of heparinized saline ranging from 19 to 63 days, serum creatinine remained normal in all but 1 animal, which developed pyelonephritis and catheter-tip perforation of the iliac artery. No cases of arterial thrombosis or catheter migration were observed at necropsy. In 7 additional autotransplanted dogs, simultaneous iliac vein and systemic (jugular vein) concentrations of 6-mercaptopurine (6-MP), the major immuno-suppressive metabolite of azathioprine, were determined during a continuous 24-hr i.a. infusion (10 mg/kg/24 hr). Following termination of the infusion, 10 mg/kg 6-MP was administered to the same 7 dogs as an i.v. bolus, and systemic drug concentrations were deter mined over a 4-hr period. Mean ±SE total-body clear ance and elimination half-life were 887 ±159 ml/min and 1.4 ± 0.2 hr, respectively, in the i.v. bolus study, indicating that 6-MP is rapidly cleared from the sys temic circulation. Unexpectedly, the kidney removed as much as 60–95% of locally infused 6-MP, reducing the amount of active drug entering the systemic circulation to 5–40% of that which would be present during an i.v. infusion of the same dose. According to the principles governing the advantages of i.a. infusions, these data demonstrate that 6-MP can be infused intrarenally to produce both a 4-fold increase in drug concentration