A study was undertaken to examine the accuracy and precision of the measurement of flow by magnetic resonance imaging (MRI) with consideration to the equipment and patient related parameters that might be encountered in vivo. For this purpose, test objects were devised consisting of PVC tubing, in which the internal diameter simulated the size of the arteries in the body. The design of the test objects ensured that steady laminar flow was obtained in the sections being imaged. The calibration study suggests that, using MRI, flow can be measured in vitro with systematic error of better than 7.0 +/- 5.0% and random error of better than 7.5%. In general, flow measurements obtained from MRI were found to correlate well with the known flows. However, the results indicated that there are prerequisite conditions for the validity of the measurements, such as the selection of appropriate flow pulse sequences and velocity limits. Measurements taken at vertically 40 mm away from the isocentre of the magnetic field were significantly different (p less than 0.01) from that at the isocentre.