Intimal hyperplasia continues to be a major problem following vascular surgery but experimental evidence suggests that it can be reduced pharmacologically. For clinical studies an accurate, reproducible and non-invasive image of the intima and lumen is required. We have assessed the value of the Acuson 128 Colour Duplex for such studies. Ten patients had their common femoral arteries scanned at a fixed point by two experienced observers on two separate occasions. External vessel diameter, luminal diameter and internal diameter (i.e. the diameter within the internal elastic lamina) were measured in both longitudinal and cross-sectional views. Cross-sectional area and degree of stenosis were all measured and all parameters expressed as limits of agreement. The mean external diameter of the common femoral arteries was 10.5 +/- 1.6 mm. Measurements in the longitudinal view were highly reproducible with limits of agreement ranging from -0.67 - +0.25 mm (internal diameter) to -1.49 - +1.31 mm (luminal diameter). In order to detect a meaningful change in longitudinal external diameter a real difference of 0.86 mm is required representing a change of less than 10%. Cross-sectional diameter measurements were similarly reproducible (-0.73 - +0.47 mm to -1.97 - +1.79 mm). However, cross-sectional area measurements had a wide variation so that the error in degree of stenosis was -25.4 - +30.2%. Thus, duplex ultrasound reproducibly images the layers of the arterial wall. Prospective studies of intimal hyperplasia are feasible but must be based on longitudinal and cross-sectional diameters rather than cross-sectional areas.