The aim of this study was to compare the influence of flow in the velocity range 0 to 25 mm/s on modulus, phase, real and imaginary images obtained with a standard magnetic resonance scanner (Siemens Magnetom, 0.5 T), and to develop a simple method for determination of flow velocities in vivo from this information. Using a flow phantom, the flow dependent magnetic resonance imaging (MRI) signal has been studied as a function of flow perpendicular to the image slice with non-doped water (simulating moving cerebrospinal fluid) as well as with water doped with Mn2+ (simulating moving blood) for each of the four mentioned image types. The results show a marked flow dependence on all types of images studied. The variation of the signal with flow in the modulus images is relaxation-time dependent in the studied velocity range and it is non-monotone for non-doped water. In the phase images, however, the variations are monotone and not dependent on relaxation times. In modulus images the curve shape is relatively independent on flow direction, while phase images are clearly dependent on flow direction in the studied velocity range. The signal versus velocity curves for the real and imaginary images show resemblance to those for the modulus and the phase images, respectively. It is concluded that the phase information can be used to generate a signal versus velocity calibration curve, which can be used to quantify low flow velocities in vivo.