In an investigation to determine the influence of sampling variability on the diagnostic yield of liver biopsy, 3 consecutive samples were obtained from each of 75 patients by redirecting the biopsy needle through a single entry site. In 14.7% of patients all 3 specimens were normal, and in 36% there were similar abnormalities in all 3 specimens. In the other patients, sampling variability between specimens was present. In those patients with cirrhosis, hepatocellular carcinoma, metastatic carcinoma, or hepatic granulomas the histological abnormality was present in all 3 biopsy specimens in only 50%, 54.5%, 50%, and 18.8% of patients, respectively. No complications were recorded. These findings show that important pathology can be overlooked if only a single biopsy specimen is taken, and that the method of obtaining 3 consecutive specimens improves the diagnostic yield of liver biopsy without an associated increase in complications.