Between 1968 and 1979, 192 aortofemoral bifurcation grafts were placed for aortoiliac occlusive disease. The proximal anastomosis was end-to-end in 101 and end-to-side in 91 patients. Anastomoses were end-to-end in 38 (37%) of 104 patients from 1968 to 1976 and 63 (72%) of 88 patients from 1977 to 1979. Embolization occurred intraoperatively in four end-to-end and five end-to-side anastomoses. The grafts became occluded in the early postoperative period in two patients with end-to-end proximal anastomoses. There were two aortoduodenal fistulas, both associated with end-to-side proximal anastomosis. The cumulative patency rate at 5 years was 87% for end-to-end and 85% for end-to-side anastomoses. An end-to-end proximal anastomosis is indicated for associated aneurysmal disease or in the presence of aortic occlusion, while an end-to-side anastomosis is indicated when there are low-lying accessory renal arteries or in the presence of occlusive disease in the external iliac arteries. The author's experience suggests that there is no difference in the incidence of intraoperative embolization or late occlusion between end-to-side and end-to-end proximal anastomosis. The incidence of aortoduodenal fistula appears to be lower with end-to-end proximal anastomosis possibly because of better tissue apposition at the anastomosis.