Studies on the complication rate of liver biopsy have hitherto been conducted in referral hospital centers. They are therefore not representative for general practice where liver biopsy is performed by specialists and nonspecialists. In a postal nationwide survey, we approached all gastroenterologists and hospital internists to assess the complication rate and practice (setting, needle type, use of ultrasonography) of percutaneous liver biopsy performed in 1989 in Switzerland for diffuse liver disease. Two hundred eighty questionnaires were mailed and 252 were returned (response rate 90.0%) 165 respondents (65.5%) performed 3501 biopsies while 87 respondents (34.5%) did not practice liver biopsy; 67.7% of biopsies were executed blindly and 32.3% were guided. Eight nonfatal and three fatal complications occurred. Hemorrhage was the most frequent complication (five cases) and was responsible for all three fatal outcomes. The overall complication rate was 0.31%, being distinctly lower in the group of gastroenterologists (0.11%) as compared to the group of internists (0.55%; P = 0.031). The complication rate was 1.68% in the group of internists performing fewer than 12 biopsies per year, while there was no complication in the group of internists performing more than 50 biopsies per year (P = 0.036). Complications were not related to the needle diameter or to the absence of ultrasonography before biopsy. In conclusion, this representative survey in Switzerland shows that the complication rate of liver biopsy is mainly related to the experience and training of the operator.