The isolated arterially perfused rabbit interventricular septum was used to determine the feasibility of using the glucose analogue 18F-2-deoxy-2-fluoro-d-glucose (DG) with a tracer kinetic model to estimate the rate of exogenous glucose utilization. FDG was delivered to the septum by constant infusion, and tissue 18F radioactivity was measured as a function of time by external coincidence counting. The following four conditions were studied: flow rates of 0.5, 1.0, and 1.5 ml/min with a heart rate of 72 beats/min and flow at 1.5 ml/min with 96 beats/min. The rate constants for FDG forward and reverse transport between the vascular and extravascular compartments (k*1, k*2, respectively), phosphorylation of FDG (k*3), and dephosphorylation of FDG-6-phosphate (FDG-6-P) (k*4) were determined from the tissue curves using a tracer kinetic model. The lumped constant (LC) of the deoxyglucose model calculated using Fick-derived myocardial metabolic rates of glucose (MMRGlc), was 0.60 +/- 0.10 and was stable over the range of conditions studied. Average k*'s and LC were used to calculate MMRGlc's employing the model and were not significantly (P greater than 0.05) different from those determined by the Fick method. Tissue analyses using high-pressure liquid chromatography documented that tissue 18F radioactivity wa due to FDG and FDG-6-P, and their relative fractions agreed well with the values predicted from the tracer kinetic model. Only FDG was detected in the effluent. These studies also indicate the presence of a myocardial enzyme that can hydrolyze FDG-6-P to FDG. Thus our results support the use of the FDG method with positron-computed tomography for the in vivo determination of the myocardial rate of exogenous glucose utilization.