Tracer techniques and quantitative autoradiographic and tissue counting models for measurement of metabolic rates were combined with positron computed tomography (PCT) and (F-18)2-fluoro-2-deoxy-D-glucose (FDG) for the measurement of local cerebral metabolic rate for glucose (LCMRGlc) in humans. A three-compartment model, which incorporates hydrolysis of FDG-6-PO4 to FDG, was developed for the measure of kinetic constants and calculation of LCMRGlc. Our model is an extension of that developed by Sokoloff et al. Although small, hydrolysis of FDG-6-PO4 was found to be significant. A PCT system, the ECAT, was used to determine the rate constants, lumped constant, and stability of the model in human beings. The data indicate that cerebral FDG-6-PO4 in humans increases for about 90 minutes, plateaus, and then slowly decreases. After 10 minutes, cerebral blood FDG activity levels were found to be a minor fraction of tissue activity. Precursor pool turnover rate, distribution volumes, and red blood cell-plasma concentration ratios were determined. Reproducibility (precision) of LCMRGlc measurements (approximate 2 cm2 regions) was +/- 5.5% over a 5-hour period. The replacement of arterial blood sampling with venous sampling was validated.