The aim of this study was to compare the diagnostic accuracy of the late phase of CEUS and the hepatobiliary phase of CE-MR with Gd-BOPTA in the characterization of focal liver lesions in terms of benignity and malignancy. A total of 147 solid focal liver lesions (38 focal nodular hyperplasias, 1 area of focal steatosis, 3 regenerative nodules, 8 adenomas, 11 cholangiocarcinomas, 36 hepatocellular carcinomas and 49 metastases) were retrospectively evaluated in a multicentre study, both with CEUS, using sulphur hexafluoride microbubbles (SonoVue, Bracco, Milan, Italy) and CE-MR, performed with Gd-BOPTA (Multihance, Bracco, Milan, Italy). All lesions thought to be malignant were cytohistologically proven, while all lesions thought to be benign were followed up. Sensitivity, specificity, positive (PPV) and negative (NPV) predictive values and accuracy were calculated for the late phase of CEUS and the hepatobiliary phase of CE-MRI, respectively, and in combination. Analysis of data revealed 42 benign and 105 malignant focal liver lesions. We postulated that all hypoechoic/hypointense lesions on the two phases were malignant. The diagnostic errors were 13/147 (8.8%) by CEUS and 12/147 (8.2%) by CE-MR. Sensitivity, specificity, PPV, NPV and accuracy of the late phase of CEUS were 90%, 93%, 97%, 80% and 91%, 93%, 97%, 81% and 92% for the hepatobiliary phase of CE-MRI, respectively. If we considered both techniques, the misdiagnosis diminished to 3/147 (2%) and sensitivity, specificity, PPV, NPV and accuracy were 98%, 98%, 99%, 95% and 98%. The combination of the late phase of CEUS and the hepatobiliary phase of CE-MR in the characterization of solid focal liver lesions in terms of benignity and malignancy is more accurate than the two techniques used separately.