In the United States of America, vascular access complications are the most frequent cause of death among patients with end-stage renal disease on hemodialysis. The majority of American patients with end-stage renal disease have prosthetic grafts for access. The most common complication of prosthetic graft fistulas is thrombosis. To prolong graft life and decrease the frequency of operations, a university-based dialysis unit adopted a policy of graft surveillance by measuring venous resistance, followed by fistulagram and percutaneous dilatation of identified stenoses. The purpose of this retrospective study is to determine if this policy affected graft patency and graft life.This retrospective review compares outcomes for two years before adoption of this policy (control group, 210 patients) with the outcomes for three years after adoption (study group, 260 patients). Fifty (24 percent) of the control group and 71 (27 percent) of the study group were defined as "complicated" patients because they underwent more than one intervention over any 12-month period.The 50 complicated patients in the control group underwent 104 operations for thrombosis over two years, 1.04 thrombotic episodes and operations per patient year. Seventy new grafts were placed with a mean primary patency of 3.9 months. Mean graft survival time was 6.3 months. The 71 complicated patients in the study group underwent 111 fistulagrams, 80 angioplasties, and 110 operations over three years, for 0.52 thrombotic episodes per patient year (p < .001). Forty-five new grafts were placed with a primary patency of 11.5 months (p < .001). Mean graft survival time was 15.8 months (p < .001).This retrospective study demonstrates the effectiveness of a policy of graft surveillance and percutaneous treatment of graft stenosis in prolonging primary surgical patency and graft survival.