OBJECTIVE To evaluate a surveillance programme based on colour duplex ultrasound scanning of above-knee prosthetic femoropopliteal bypass. METHODS A retrospective clinical study. METHODS One hundred and eighteen above-knee prosthetic femoropopliteal bypass procedures performed from 1993 to 1996. METHODS Scans were done at 1, 3, 6 and 12 months, and annually thereafter. Significant graft-related stenotic lesions were defined as 50% diameter reduction or more located at the inflow or outflow arteries, or associated with the graft. Identified lesions were treated by angioplasty or surgical repair. RESULTS The primary and secondary patency rates at 2 years were 55% and 71%, respectively. Stenotic lesions were documented on 45 of 463 scans (9.7%). Twenty-nine grafts (25%) were treated and only six of these grafts (21%) occluded. This was significantly superior to the occlusion rate of 41% for the whole series. CONCLUSIONS It appears that one-quarter of above-knee prosthetic femoropopliteal bypasses develop graft-related stenoses. The favourable prognosis of these grafts after treatment for stenotic lesions justifies a surveillance programme. However, most graft occlusions seem to occur in grafts where no stenotic lesion has been previously detected.