The deoxyglucose method originally developed for measurements of the local cerebral metabolic rate for glucose has been investigated in terms of its application to cardiac studies with positron computed tomography (PCT) and fluorodeoxyglucose (FDG). Studies were performed in dogs to measure the tissue kinetics of FDG with PCT and by arterial and venous sampling. The operational equation developed in our laboratory as an extension of the Sokoloff model was used to analyze the data. Error propagation, primarily from corrections applied to remove spillover of activity from the myocardial blood pool to tissue and from partial-volume effects in the PCT images, limited accuracy in the estimation of the individual rate constants for transport, phosphorylation, and dephosphorylation. However, a constant representing the combination of transport and phosphorylation was accurately determined and yielded measured values of the myocardial metabolic rate for glucose (MMRGlc) that were in good agreement with direct determinations using the Fick method over a wide range of glucose metabolic rates (from 1.7 to 21.1 mg/min-100 g). The lumped constant (0.67 +/- 0.10) was also found accurate and stable over this range of metabolism. The FDG method accurately predicted the true MMRGlc even when the glucose metabolic rate was normal but myocardial blood flow (MBF) was five times the control value, or when metabolism was reduced to 10% of normal and MBF increased to five times normal. Improvements of PCT resolution are required to improve the accuracy of the estimates of the rate constant and the MMRGlc.