The model for quantifying local cerebral glucose metabolic rates originally developed by Sokoloff et al. and modified by Phelps. Huang and co-workers was applied to humans with cerebral ischemia (i.e., stroke). Rate constants for fluorodeoxyglucose were measured in ischemic and nonischemic regions with position computed tomography. Using measured rate constants for ischemia, the model generate more accurate estimates of local cerebral glucose metabolism as compared to the use constants from normal young adults, because the local metabolic rate is significantly underestimated, and temporal instability of the model is observed when normal values are applied to ischemic regions. A method was also developed to test the stability of the local lumped constant. The estimates of the lumped constant showed no or only small variations between ischemic and nonischemic types. Thus, errors introduced in the calculated local cerebral glucose metabolism by inappropriate rate constants appear to be more significant than those caused by any potential change in the lumped constant in ischemia.