Ultrasound measurement of volume blood flow is potentially useful for many clinical situations, yet practical implementation and use are restricted by the many instrumentation and blood flow limitations that can arise. Colour velocity imaging offers a number of theoretical advantages over methods based on duplex imaging. We evaluated a colour velocity flow measurement system (CVI-Q, Philips) both in a flow phantom and in vivo in the extracranial carotid arteries of normal volunteers. Over a range of constant (50-1200 ml/min) and pulsatile (92-366 ml/min) flows and using both steered and unsteered beams with beam angles of 30 degrees and 40 degrees, errors usually within 5% were obtained for constant flow and within 10% for pulsatile flow. However, with a beam angle of 70 degrees, higher errors of 20% were obtained for pulsatile flow. The reproducibility of flow measurements made using both anterior and posterior-lateral scanning approaches was determined in the common (CCA), internal (ICA) and external carotid (ECA) arteries of 18 volunteers. A greater reproducibility was found using the posterior-lateral approach (CCA 6.27%; ICA 9.8%), and mean (SD) flow values were 376 ml/min in the CCA and 255 ml/min in the ICA. The ratio of (ICA + ECA)/CCA flow calculated for each subject individually was mean (SD) 0.95 (0.11). Insonation from an anterior approach resulted in lower reproducibility and lower flow values. In conclusion, colour flow velocity imaging allows repeatable reproducible measurements of CCA and ICA flow, but results are optimal if a posterior-lateral scanning approach is used.