The equilibrium C15O2 inhalation method for measuring cerebral blood flow with positron emission tomography (PET) is based on a one-compartment model for which it is assumed that the local flow and partition coefficient are uniform in the tissue region in which flow is to be determined. However, because of the limited spatial resolution of PET, a region of interest will contain a mixture of gray and white matter. We used a computer simulation to examine the effect of this heterogeneity on flow measurement in both normal and pathological states. With gray and white matter flows of 0.80 and 0.20 ml/min/g, respectively, flow is underestimated by a maximum of 20% in a region that is 30% gray. Errors occur not only because of flow heterogeneity, but also because of heterogeneity of partition coefficient and the sensitivity of the method to errors in partition coefficient. Larger errors occur in the case of cerebral hyperemia, although the method becomes more accurate with ischemia. In the case of simulated brain tumor, the accuracy of flow determination varies considerably, depending on the flow and partition coefficient of the tumor and of the surrounding tissue. Finally, incremental changes in gray matter flow, as would occur with functional cortical activation, are not well reflected. Thus, the equilibrium C15O2 inhalation method is limited in its ability to accurately quantitate local cerebral blood flow in heterogeneous tissue regions.