Experimental transplantations of cadaver bone marrow (BMT) in beagle dogs were performed to evaluate the problems and potentials in a preclinical setting. The effectiveness of selective decontamination of the gut (SD) and gnotobiotic surveillance in preventing infections during longer aplastic periods was investigated. Three groups of dogs were compared. Group A: controls. Group B: dogs with BMT, without SD and irregular gnotobiotic surveillance. Group C: dogs with BMT, with SD and regular gnotobiotic surveillance. The intestinal colonization of normal healthy beagles shows similarities as well as dissimilarities to the human intestinal microflora. Aerobic potentially pathogenic organisms do not colonize the gut of healthy beagles under our keeping conditions. SD resulted in a significant decrease in infections with Escherichia coli and Plesiomonas. Infections with anaerobes, as well as bacterial infections of the respiratory tract were, however, not prevented. The intestinal colonization in dogs of group C with Clostridium difficile is another obvious effect of SD. The infections encountered during the study indicate the importance of the "take" for the clinical significance and outcome of intestinal colonization with potentially pathogenic organisms. In order to reduce the drug burden of BMT patients, we consider the elimination of routine SD after BMT not to be superior to gnotobiotic surveillance and germ-specific short term decontamination.