Streamline display is a unique alternative to cross-sectional slice or projection display, because streamlines more clearly show the patterns of blood flow within the vessel. Flow patterns associated with atherosclerosis, such as streamline separation and recirculation, can be quickly identified with this display. Streamlines can be calculated using velocity data obtained from 3-D phase contrast angiographic pulse sequences. However, these streamlines often pass through the wall of vessel or show intraluminal sources and sinks of blood. The author has developed iterative least squares algorithms to improve the realism of streamlines. The velocity data is modified so that the resulting streamlines do not pass through the vessel wall and there are no intraluminal sources or sinks. He has applied the algorithms to velocity data obtained from a flow phantom and the carotid arteries of normal volunteers. Streamlines derived from the processed velocity fields are more realistic and provide more precise flow quantitation.