Twenty-two patients with primary or metastatic brain tumors were evaluated with computerized tomography (CT) and intraoperative ultrasound. Tumor volume was estimated using a geometric formula based on CT and intraoperative ultrasound measurements. In most cases, tumor margins were marked with indigo carmine injected by ventricular cannula or with Silastic ventricular catheters placed under ultrasonographic control. Nine tumors had previously been operated on and irradiated (Group A). In this group, intraoperative ultrasound tended to overestimate the tumor volume compared to CT (intraoperative ultrasound findings 141.39% +/- 37.73% of CT findings (mean +/- standard deviation]. Sixteen patients were operated on for the first time (Group B), and in this group the volume estimates were comparable (intraoperative ultrasound findings 101.69% +/- 24.65% of CT findings). The difference between the means was statistically significant (p less than 0.01). Ultrasonography improved intraoperative delineation of tumor margins, as depicted by CT, thus maximizing the extent of resection; however, with recurrent tumors, intraoperative ultrasound tended to overestimate the tumor volume. Gliosis may account for this difference.