Three-dimensional (3D) sonography represents a further development of non-invasive, diagnostic sonographic imaging techniques. The clinical use of this technique in the hepatobiliary system reveals its possible applications, especially in the case of localized lesions. The goal of this study was to evaluate the accuracy of in vivo, three-dimensional ultrasonography with a freely moveable transducer as compared with a computerized tomographic arterial portography using the spiral technique (S-CTAP). A total of 42 patients with localized hepatic lesions (1.5 to 1231 cm3, mean 155 cm3) standard deviation (SD) 231 cm3 were examined using conventional (2D) and 3D ultrasonography (US). Eleven of these patients additionally underwent spiral CTAP. The volumes were calculated using the ellipsoid formula (2D US and 3D US) and were also reconstructed planimetrically with the aid of work stations (3D US and 3D CT). Furthermore, the intra and inter-investigator variability of the 3D planimetry was determined. The conformity of US methods with those using S-CTAP was determined with Bland and Altman s limits of agreement. Inter-rater agreement was calculated using Cohen s concordance index kappa. The deviation of 2D US from 3D CT was -62% to +68% and that of 3D US (ellipsoid) from 3D CT was -28% to +9%. Between 3D US (planimetry) and S-CTAP the deviation was -21% to 9%. The kappa value was 0.886. The intra-investigator variability was 5%. Our results show that the planimetric 3D US is independent of the investigator and, with regard to the volume determinations, is substantially more accurate than measurements with conventional sonography and comparable with those measurements made using CT investigations. 3D sonography may be applied complementarily to CT as an economical procedure for the follow-up of tumor disease.