We prospectively evaluated the clinical efficacy of ultrasonography (US) in monitoring the effect of medical treatment in patients with liver metastases, by comparing serial US assessment with serial magnetic resonance imaging (MRI) evaluation and clinical outcome in a group of 41 patients with solid tumors. Both examinations were performed in patients with metastatic liver disease at the start of a new treatment modality and monthly thereafter for 3 months; close monitoring was prolonged beyond the third month in cases in which there was disagreement between the two techniques and the clinical course was not conclusive. Planned follow-up was completed in 37 cases. There was limited concordance between the two examinations: in 21 cases only (56.8%), US and MRI gave concordant information on the evolution of hepatic metastases; in eight cases, both agreed on progression of disease (PD), in 11 cases on stable disease (SD), and in one case each on partial response (PR) and complete response (CR). In the remaining 16 cases (43.2%), there was disagreement between the two examinations. On the basis of subsequent clinical course, this discrepancy was shown to be due to US inadequacy in 13 cases and to MRI inadequacy in one case; in two cases, the clinical course was not conclusive. The most striking limits of US appeared to be twofold: (1) a progressive appearance, with chemotherapy, of a diffusely inhomogeneous structure of the liver, resulting in obscuration of focal lesions (and a subsequent judgement of CR) in cases in which lesions were, on the contrary, detected at MRI and usually confirmed by subsequent clinical course; and (2) false US-determined PD in cases in which lesions proven at baseline MRI were noted at US only after one to two courses of therapy. We conclude that US, which is known to be inaccurate for screening of liver metastases, is unreliable for the follow-up of metastatic liver disease; despite its wide availability, low cost, and noninvasiveness, critical therapeutic decisions should not be made based on the outcome of this test.