This study was designed to compare diagnostic accuracies of measuring liver enzymes, preoperative ultrasonography, surgical examination, and intraoperative ultrasonography for detection of liver metastases from colorectal cancer. A blind prospective comparison between the diagnostic examinations mentioned above were performed in 295 consecutive patients with colorectal cancer. An experienced ultrasonologist performed the preoperative examinations and the results were not known to the other experienced ultrasonologist, who did the intraoperative examinations. The latter was also unaware of the findings by the surgeon. The presence of metastases was further assessed by ultrasonography three months postoperatively, as well as surgery and liver biopsy in some of the patients. The sensitivity of intraoperative ultrasonography (62/64) was significantly superior to that of surgical exploration (54/64), and that of preoperative ultrasonography (45/64). The lowest sensitivity was presented by liver enzymes. "Bilobar" metastases were detected in 42 of 46 patients by intraoperative ultrasonography, but in no more than 33 by the surgeon. Intraoperative ultrasonography demonstrated the highest specificity of all examinations. Intraoperative ultrasonography reduces the number of patients with liver metastases being subjected to superfluous or even harmful liver surgery and it may increase the number in whom liver surgery will prolong life.