Detre et al. (Magn. Reson. Med. 23, 37-45 (1992)) and Zhang et al., (Magn. Reson. Med. 25, 362 (1992)) have recently demonstrated a technique for the measurement of regional cerebral blood flow (rCBF) based on the continuous saturation (or inversion) of the arterial blood supply to the brain in rats at 4.7 T. In the work reported here, we combined this technique with volume localized (PRESS) readouts to benefit from recording "perfusion" signals averaged over a larger volume, resulting in rapid acquisition of data with sufficient signal-to-noise ratio for application at 2.0 T. In 10 baseline flow measurements, the mean error between the NMR technique and the microsphere flow measurement was -1.5% with a standard deviation of 15.2%. For five measurements obtained with occlusion of the middle cerebral artery, the mean error was -32.4 +/- 20.2%. Perfusion measurements from a single animal under hypercapnic conditions indicated that the NMR technique could underestimate rCBF at high flow rates. An error analysis of the NMR perfusion model is also presented, along with results for typical parameters encountered at 2.0 T.