One hundred twenty-five patients investigated at the Royal Brisbane Hospital, who underwent both hepatic ultrasonography and liver biopsy between 1980 and 1983, were scored quantitatively for ultrasound features of loss of detail, echogenicity, and attenuation, as well as for histologic features of fat, fibrosis, and inflammation. Strong correlations were found between the score for fat content and each of the three ultrasound features, and between the score for hepatic fibrosis and loss of detail and echogenicity, but there was no strong correlation with attenuation. Hepatic inflammation did not correlate with any of the ultrasound features. The correlations for fat were strongest when the interval between ultrasonography and liver biopsy was less than or equal to 7 days. Although ultrasonography had a positive predictive value of 98% in the diagnosis of diffuse parenchymal abnormality, it did not distinguish fat from fibrosis nor reliably diagnose cirrhosis. Ultrasonography gave false-positive results in only 2 patients, but in 17 patients, false-negative ultrasound examinations were encountered. These findings indicate that ultrasound is not a useful screening investigation for parenchymal liver disease, nor is it useful in gauging hepatic pathology. However, abnormal hepatic ultrasonography in patients with suspected liver disease strongly suggests the presence of diffuse liver disease.