On the basis of the assumption that oxygen delivery across the endothelium is proportional to capillary plasma PO2, a model is presented that links cerebral metabolic rate of oxygen utilization (CMRO2) to cerebral blood flow (CBF) through an effective diffusivity for oxygen (D) of the capillary bed. On the basis of in vivo evidence that the oxygen diffusivity properties of the capillary bed may be altered by changes in capillary PO2, hematocrit, and/or blood volume, the model allows changes in D with changes in CBF. Choice in the model of the appropriate ratio of Omega identical with (DeltaD/D)/(DeltaCBF/CBF) determines the dependence of tissue oxygen delivery on perfusion. Buxton and Frank (J. Cereb. Blood Flow. Metab. 17: 64-72, 1997) recently presented a limiting case of the present model in which Omega = 0. In contrast to the trends predicted by the model of Buxton and Frank, in the current model when Omega > 0, the proportionality between changes in CBF and CMRO2 becomes more linear, and similar degrees of proportionality can exist at different basal values of oxygen extraction fraction. The model is able to fit the observed proportionalities between CBF and CMRO2 for a large range of physiological data. Although the model does not validate any particular observed proportionality between CBF and CMRO2, generally values of (DeltaCMRO2/CMRO2)/(DeltaCBF/CBF) close to unity have been observed across ranges of graded anesthesia in rats and humans and for particular functional activations in humans. The model's capacity to fit the wide range of data indicates that the oxygen diffusivity properties of the capillary bed, which can be modified in relation to perfusion, play an important role in regulating cerebral oxygen delivery in vivo.