In this prospective trial, intraoperative 2-dimensional (2-D) and 3-dimensional (3-D) transesophageal echocardiography (TEE) examinations were performed on 60 consecutive patients undergoing cardiac valve surgery. Both 2-D (including color flow and Doppler data) and 3-D images were reviewed by blinded observers, and major valvular morphologic findings recorded. In vivo morphologic findings were noted by the surgeon and all explanted valves underwent detailed pathologic examination. To test reproducibility, 6 patients also underwent 3-D TEE 1 day before surgery. A total of 132 of 145 attempted acquisitions (91%) were completed with a mean acquisition time of 2.8 +/- 0.2 minutes. Acquisition time was significantly shorter in patients with regular rhythms. Reconstructions were completed in 121 of 132 scans (92%) and there was at least 1 good reconstruction in 56 of 60 patients (93%). Mean reconstruction time was 8.6 +/- 0.7 minutes. Mean effective 3-D time, which was the time taken to complete an acquisition and a clinically interpretable reconstruction, was 12.2 +/- 0.8 minutes. Intraoperative 3-D echocardiography was clinically feasible in 52 patients (87%). Three-D echocardiography detected most of the major valvular morphologic abnormalities, particularly leaflet perforations, fenestrations, and masses, confirmed on pathologic examination. Three-D echocardiography predicted all salient pathologic findings in 47 patients (84%) with good quality images. In addition, in 15 patients (25%), 3-D echocardiography provided new additional information not provided by 2-D echocardiography, and in 1 case, 3-D echocardiographic findings resulted in a surgeon's decision to perform valve repair rather than replacement. In several instances, 3-D echocardiography provided complementary morphologic information that explained the mechanism of abnormalities seen on 2-D and color flow imaging. In the reproducibility subset, preoperative and intraoperative 3-D imaging detected a similar number of findings when compared with pathology. Thus, in routine clinical intraoperative settings, 3-dimensional TEE is feasible, accurately predicts valve morphology, and provides additional and complementary valvular morphologic information compared with conventional 2-D TEE, and is probably reproducible.