The [(18)F]fluorodeoxyglucose (FDG) method to measure glucose metabolism quantitatively in humans is reviewed. The assumptions and the mathematical formulation of the underlying autoradiographic Sokoloff model and its adaptation to positron emission tomography (PET) are described. Various implementations to estimate glucose consumption from measured tissue activity with PET are presented. The dependence on the "lumped constant" and on the accuracy of the input function is discussed. Recommendations for the practical application of different procedures for performing FDG studies are given.