A prospective audit of 644 patients undergoing biliary tract operations has been conducted to assess the incidence of bile colonization and its association with the incidence of postoperative sepsis when all patients received the same prophylactic antibiotic. The accuracy of the determination of high-risk factors has been assessed as has the correlation between bile colonization and patients assessed as 'high risk'. Organisms were cultured from the bile of 121 (19 per cent) patients and among these the incidence of wound or intra-abdominal sepsis was 22 per cent whereas among patients with sterile bile the incidence was only 2 per cent (P less than 0.0001). Although the incidence of bile colonization within the high-risk group (32 per cent) was more than twice that in the low-risk group (14 per cent), more than half (54 per cent) of the patients with positive bile cultures were in the low-risk group. It is concluded that, despite prophylactic antibiotics, bile colonization remains the major factor associated with postoperative sepsis, but that this cannot be predicted accurately by preoperative assessment of high-risk factors. Furthermore, we believe that a policy of selective administration of prophylactic antibiotics solely to high-risk patients cannot be justified.