OBJECTIVE To compare axial reconstructed ultrasonographic images from three-dimensional data volumes and conventional computed tomographic scans in documentation and detection of liver lesions. METHODS The livers of 23 patients were examined by a multifrequency curved array transducer to acquire three-dimensional data volumes and by conventional computed tomography The ultrasonographic device was equipped with three-dimensional ultrasonographic technology allowing for real-time parallel movement in every plane within an acquired data volume. Axial ultrasonographic images reconstructed from three-dimensional data volumes were compared with conventional computed tomographic scans. RESULTS When both methods were taken together, a total of 51 different liver lesions could be detected. Reconstructed ultrasonographic images depicted 44 (86%) of 51 lesions, and computed tomographic scans showed 46 (90%) of 51 lesions. Compared with computed tomography, ultrasonography depicted 5 additional lesions in 3 of the patients (4 hemagiomas and 1 unspecified lesion), whereas 7 lesions were missed in another 4 patients (4 metastases, 2 calcifications, and 1 cyst). The Pearson correlation coefficient between ultrasonography and computed tomography was r = 0.84 (P < .001). CONCLUSIONS Our study shows that ultrasonographic images can be effectively reconstructed from three-dimensional data volumes. With respect to documentation and detection of liver lesions, the results obtained by three-dimensional ultrasonography appear comparable with those obtained by conventional computed tomography. However, several technical and procedural limitations have to be respected.