This report describes a strategy for measurement of regional CBF that rigorously accounts for differing tracer partition coefficients and recirculation, and is convenient for use with positron emission tomography. Based on the Kety model, the measured tissue concentration can be expressed in terms of the arterial concentration, the rate constant K, and the blood flow f. The local partition coefficient may be computed as p = f/K. In our approach, maps of K and f are computed from two transverse section reconstructions. The reconstructions are based on weighted sums of projection data measured frequently during the observation period. Theoretical studies of noise propagation in the estimates of K and f were carried out as a function of tomographic count rate, total measurement time, and tracer half-life for varying input functions. These calculations predict that statistical errors in f of between 5 and 10% at a resolution of 1 cm full width at half maximum can be obtained with existing tomographs following i.v. injection. To compare theory and experiment, a series of flow studies were carried out in phantoms using a positron tomograph. These measurements demonstrate close agreement between computed flow and noise estimates and those measured in a controlled situation. This close agreement between theory and experiment as well as the low statistical errors observed suggest that this approach may be a useful tool in clinical investigation.