Values of regional cerebral oxygen extraction ratio and oxygen utilisation obtained with the oxygen-15 steady-state inhalation technique have been found to be overestimated due to the signal from intravascular oxygen-15. A previously described method to correct for this intravascular component has been applied to a series of studies on normal subjects, and on brain tumour and stroke patients. With this correction the regional cerebral oxygen extraction ratio in normals becomes comparable to the global values previously reported with arteriovenous sampling techniques. Within the lesions of brain tumour and stroke patients, the corrections have been found to be variable and often substantial. It is concluded that failure to apply this correction may result in major errors in the values for regional oxygen extraction ratio and oxygen utilisation. This is especially true when the regional blood flow and oxygen extraction ratio of a tissue is low and regional blood volume is high.