To assess the accuracy of volumetric measurement of high-grade intracranial gliomas by computed tomography (CT), the scans of 26 patients were studied retrospectively. Twenty patients had only pre-treatment scans while six had scans both before and after chemotherapeutic treatment. Tumour volumes were measured separately by two observers, each on two occasions. A significant intraobserver variation in volumetric measurement was encountered, as well as a systematic variation between the observers; one radiologist consistently reading higher than the other. The after treatment readings were more variable than the pre-treatment readings, and most changes in tumour volume following treatment were obscured by the intra-observer variation. This study provides further evidence of the limitations of volumetric measurement as a means of assessing the response of high-grade gliomas to treatment.