A problem of current MRA techniques is the inability to accurately depict the vascular anatomy, particularly in areas of disturbed flow. Various reasons, such as intravoxel phase dispersion, saturation, temporal variations, and maximum intensity projection (MIP) nonlinearity, cause a wrong delineation of vessel boundaries. A phase contrast (PC)-based postprocessing operation, the phase derivative (PhD), is introduced to detect phase fluctuations indicating flow. Two-dimensional and three-dimensional angiographic reconstruction algorithms are presented. Mathematical formulas are derived to predict the effect of sampling to flow profiles and the effect on the PhD of these profiles. Numerical, phantom, and preliminary in vivo experiments demonstrate that PhD images do not suffer from phase wraps and allow a velocity dynamic range extension only limited by a differential phase change. It is also shown that PhD MIPs produce higher signal-to-noise ratios than conventional PC angiograms and give a better impression of the anatomy of (stenotic) vessels and of their diameters for both laminar and disturbed flow.