The purpose of this study was to identify patients who were more likely to experience septicemia after endoscopic biliary drainage. In an attempt to determine the relative importance of each risk factor and their possible interdependancy to more precisely identify high-risk patients and to deduce some guidelines for prevention, a discriminant regression analysis of risk factors for septicemia was used. Clinical, biological, and radiological data of 34 consecutive patients who experienced septicemia within 3 days after endoscopic biliary stenting were reviewed retrospectively and compared with data of a group of 71 patients without any septic complication. If only data available before the procedure were used in the discriminant analysis, prior cholangitis and leucocytosis appeared as significant risk factors, but the linear combination of these data could not predict septicemia in 50% of cases. When information concerning the quality of drainage after the procedure was introduced into the analysis, 91% of the septicemic patients were identified, and other expected risk factors such as the nature of the stricture, the type of drainage, or prior cholangitis and leukocytosis had no or marginal predictive values. Patients referred from centers where duodenoscopes might have been poorly disinfected appeared to be at higher risk for Pseudomonas aeruginosa septicemia. These results emphasize the crucial role of the quality of drainage as a risk for septicemia. Regarding the prevention of infection, it is concluded from this study that (a) pure diagnostic endoscopic retrograde cholangiopancreatography should be avoided in obstructed patients if drainage cannot be performed during the same procedure; (b) drainage should be as complete as possible; (c) antibiotics should be administered before ERCP to every patient with suspected obstructive jaundice and should cover P. aeruginosa if local epidemiological data suggest that there is a problem with disinfection of the endoscopes; and (d) the quality of drainage should guide the duration of antibiotic prophylaxis.