Bleeding time determination is not advised as a general preoperative hemostasis screening test, but it might be useful in some patient groups. Patients referred for liver biopsy frequently have coagulation disturbances and are at risk of hemorrhage. In this prospective study 219 liver biopsies were carried out regardless of a prolonged bleeding time, but with minimum requirements for hemoglobin concentration, platelet count, and tests of the internal and external coagulation pathways. The bleeding time was prolonged in the case of 48 (22%) of the biopsies. Significant bleeding as defined by a hemoglobin decrease of > or =2.0 g/dl occurred in nine patients. Three of these patients were bone marrow transplanted. Patients with a prolonged bleeding time carried a five times higher risk of bleeding (odds ratio = 5.0; confidence interval = 1.1-21.8; p = 0.019). We conclude that the bleeding time may give additional information on the risk of bleeding in some patient groups undergoing liver biopsy.