Flow velocity imaging studies have been conducted by means of a selective saturation-recovery spin echo technique, and the dependence of signal amplitude on interpulse interval, echo delay, slice-selection gradient, and flow velocity was evaluated experimentally. The simple theory predicting a steady increase of signal intensity with increasing interpulse interval until this latter equals the transit time could be verified in phantoms and was shown to permit measurement of blood flow velocity in venous structures such as the femoral vein. The flow phantom experiments further showed that the final intensity, attained when inversion time (TI) = transit time, decreases with increasing flow velocity, an effect that cannot be explained by influx of spins between the 90° detection pulse and the 180° refocusing pulse. This signal reduction is due to slice-selection gradient-induced phase shifts across the pixel, caused by the intralumenal velocity gradient, leading to destructive interference of the spin isochromats. The velocity distribution can be mapped by plotting signal intensity as a function of interpulse interval for pixels in different radial positions. To highlight arterial flow, gating is required with the acquisition delay selected such that the interpulse period TI falls in a time zone of slow flow within the cardiac cycle. By subtracting images recorded with different acquisition delays, flow images showing arterial enhancement only can be obtained, as illustrated for the femoral artery in the thigh.