A postal questionnaire survey was conducted to establish the pattern of current practice of liver biopsy, to estimate the frequency of adverse events and their possible associations, and to compare respondents' personal preferences with what they offer patients. Three hundred and sixty-four members of the British Society of Gastroenterology (BSG) were questioned. There were 296 returns, of which 278 were appropriate, giving an overall response rate of 81.3%. Over the previous 2 years complications were reported in 80 patients who required blood transfusion, 11 who needed laparotomy (including 1 for pain), 13 with septicemic shock, 29 who had peritonitis, and 1 with a subphrenic collection; 53 had pulmonary complications (hemoptysis, pneumothorax and 1 hemothorax), and there were 19 deaths. The overall complication rate was 5.6 per 1,000, and the mortality rate 5.2 per 10,000. The complication rate gradually decreased with increasing number of biopsies performed until the number reached 20 per month, after which there was a significant increase in complications (chi 2 = 42.98, p < 0.001). Although 50% of the respondents were themselves prepared to have a liver biopsy on a day case basis, only 11% do it routinely on patients, and 72% do not offer it at all (chi 2 = 76.8, p < 0.001). There were no differences in complication and death rates between the above groups, but those who do occasional (< 50%) day case biopsies have a significantly higher complication rate (chi 2 = 24.3, p < 0.001).(ABSTRACT TRUNCATED AT 250 WORDS)