A consecutive series of I00 patients affected by gastrointestinal malignancies entered a prospective controlled study of liver metastases performed by ultrasound echography, CEA, hepatic enzymes (only alkaline phosphatase (AP) was found to be somehow significant). Laparotomic inspection and palpation were taken as objective control of ultrasound scan. Eighteen out of the 100 patients showed diffuse hepatic metastases at surgery (all controlled histologically). Hepatic echography correctly diagnosed liver metastases in 15 out of these 18 patients (= 83.8% sensitivity); two more cases (hepatic fibroangiomas) were interpreted as metastases ( = 89.9% specificity). CEA-RIA assay was pathologic (> 10 ng/ml) in all of the 18 patients with liver metastases; 21 out of the 82 without liver metastases were CEA positive (difference of mean values statistically significant at P < 0.01). The only significant hepatic enzyme was AP, which was pathologic in 12 out of 18 patients with liver metastases. The comparative evaluation of the three tests showed that ultrasound scanning missed three cases of diffuse hepatic metastases, which, however, were CEA positive, while AP could not help to correct such misdiagnosis.