In this study we compared the ultrasound findings of 203 hospitalized patients with a variety of reference methods: biopsy, computed tomography and laboratory liver function tests with the aim of defining their clinical relevance. The ultrasound findings were assignable to 3 groups: normal, descriptive and definitive. When ultrasound described a liver as normal, or showing "increased echogenicity" or "altered configuration", the figures of normal clinical reference methods were almost identical (i.e., 70% normal). When a definitive ultrasound diagnosis ("cirrhosis", "fatty liver" or "cardiac congestion") was made, the percentage of otherwise normal livers decreased to less than 20% and was 0% for cirrhosis and cardiac congestion. The positive predictive value for a single abnormal criterion in ultrasound was between 16% and 21%, while for a definitive diagnosis it was between 67% and 100%. Many of our patients, however, had additional risk factors for liver abnormalities, such as obesity, diabetes mellitus or chemotherapy for malignancies. These risk factors can induce morphological parenchymal alterations without blood test abnormalities and, although correctly diagnosed by ultrasound, elude other reference methods in patients without biopsy. In conclusion, the finding of a single abnormal criterion of liver abnormality in ultrasound should be treated with caution. Ultrasound diagnoses of "fatty liver", "cirrhosis", diagnosed by additional signs of portal hypertension, or "cardiac congestion", yield more information. A normal ultrasound does not exclude the presence of fatty liver or cirrhosis.