BACKGROUND: The objective of the study was to test the ability of a 3D ultrasound (US) based intraoperative imaging and navigation system to delineate gliomas and metastases in a clinical setting. The 3D US data is displayed as reformatted 2D image slices. The quality of the displayed 3D data is affected both by the resolution of the acquired data and the reformatting process. In order to investigate whether or not 3D US could be used for reliable guidance in tumour surgery, a study was initiated to compare interpretations of imaged biopsy sites with histopathology. The system also enabled concomitant comparison of navigated preoperative MR with histopathology. METHOD: Eighty-five biopsies were sampled between 2-7 mm from the tumour border visible in the ultrasound images. Biopsies were collected from 28 operations (7 low-grade astrocytomas, 8 anaplastic astrocytomas, 7 glioblastomas and 6 metastases). Corresponding cross-sections of preoperative MR T1, MR T2 and intraoperative US were concomitantly displayed, steered by the biopsy forceps equipped with a positioning sensor. The surgeons' interpretation of the images at the electronically indicated biopsy sites were compared with the histopathology of the samples. FINDINGS: The ultrasound findings were in agreement with histopathology in 74% (n = 31) for low-grade astrocytomas, 83% (n = 18) for anaplastic astrocytomas, 77% (n = 26) for glioblastomas and 100% (n = 10) for metastases. Excluding irradiated patients, the results for glioblastomas improved to 80% concurrence (n = 20). As expected tumour cells were found in biopsies outside the US visible tumour border, especially in low-grade gliomas. Navigated 3D US have a significantly better agreement with histopathology than navigated MR T1 for low-grade astrocytomas. CONCLUSION: Reformatted images from 3D US volumes give a good delineation of metastases and the solid part of gliomas before starting the resection. Navigated 3D US is at least as reliable as navigated 3D MR to delineate gliomas and metastases.