An in vivo autoradiographic strategy is described for the measurement of local cerebral blood flow in humans by positron emission tomography, based on an application of the single-compartment model originally proposed by Kety. A variety of factors are considered upon which the successful quantitation of local blood flow depends. These factors include the mode of tracer administration and the definition of the arterial input function; the choice of scan parameters to assure unique and sensitive values of flow throughout the physiological range of interest; the influence of these parameters on the stability and signal/noise characteristics of the computed flow; the error introduced by the presence of heterogeneity of flow within a volume element; and factors related to the choice of the radiotracer itself. The in vivo autoradiographic method is compared to an alternative local cerebral blood flow method employing continuous inhalation of oxygen-15-labeled carbon dioxide. The general relevance of these issues to all local blood flow methods intended for emission tomographic application is emphasized.