In an effort to define more precisely and objectively computerized tomography (CT) brain scan evidence of glioma patient response to treatment, planimetric measurements of serial CT images of enhancing tumor areas were made using a digitizing tablet interfaced to a microcomputer for computing three-dimensional tumor volumes. The ability of a single investigator to measure a 'significant change' in tumor volume was determined from that investigator's coefficient of variation (COV) for triplicate volume measurements (a total of 1701) on 155 scans of 27 patients with malignant gliomas. Planimetric volume data were compared with geometric computation of volumes based upon the product of the maximum diameter of enhancing tumor and the perpendicular diameter for each image made simultaneously with each planimetric measurement. The planimetric method COV was less than that for geometric computation, and the former method was employed for analysis of response to therapy in these same patients. Overall, for a tumor volume change to be significant (COV plus 2 standard errors of the means), the percentage change was determined to be 20%. However, the smaller the tumor volume being measured, the greater was the percentage change required in order to be significant. Thus, minimal measurable changes (%) were separately defined for large (>14 cc), medium (8 to 14 cc), and small (<8 cc) tumor volumes. Tumor volumes computed from baseline (prior to investigational therapy) and from subsequent serial CT scans were compared, with response defined as a significant change. Responses to therapy based on significant volume changes were compared in each instance to the conventional visual viewbox comparison ('Gestalt') of serial scans. In 28% of scan comparisons, planimetric technique sensitivity permitted determination of significant enlargement or reduction in tumor size, while Gestalt comparison suggested no change. The use of quantitative tumor volume analysis of planimetric determinations of changes in tumor size during investigational therapy appears to permit recognition of either progression or regression of tumor size earlier than by Gestalt comparison in one-fourth of instances.