Basic experiments were done to examine the influence of the extent of curve, direction of flow, vascular diameter, velocity of flow, and course of vessels on MR angiography. In the first experiment, two phantoms were constructed of vinyl tubing, a hairpin type and one with tubes bent to varying degrees. These phantoms were imaged in two matrices, different FOV, and of various velocities with a 1.5 Tesla system. Images of a normal volunteer were also obtained. In studies of the hairpin phantoms, a low intensity artifact appeared at the curve position, and the edge became fuzzy when the flow was perpendicular to the phase encoding direction. These phenomena were more apparent with a smaller matrix, smaller FOV, more gentle angle, larger diameter, and faster flow; in other words the component along the direction was dominant. In the phantom that used bent tubes, another low intensity artifact appeared on the medial side of the outflow portion. The results obtained from the volunteer corresponded well to those from the basic experiments. It was thought that the profile of flow was not well demonstrated due to the coarse matrix. In the second experiment, water flowed at various speeds through three-dimensional phantoms that were made of vinyl tubes of various diameters and directions. The phantoms were imaged with two- and three-dimensional time of flight and phase contrast angiography. The lower limits of detection of the phase contrast method were almost the same. In the coronal plane, the lower limit decreased to 2.5 mm for arteries and 9.5 mm for veins.(ABSTRACT TRUNCATED AT 250 WORDS)