In the 1960s, almost all patients who developed gram-negative bacteremia during granulocytopenia died; death occurred before blood culture results were available in about 50% of cases; many patients received antibiotics that were, at best, suboptimal and frequently inactive against the invading pathogen. In the early 1970s epidemiological studies demonstrated that more than 50% of gram-negative bacteremias were caused by hospital-acquired strains which colonized along the alimentary canal and caused infection in a limited number of locations, especially the pharynx, lungs, colon, and perianum. Surveillance culture studies have demonstrated that among acquired gram-negative bacilli, Pseudomonas aeruginosa will almost invariably proceed to bacteremia if the patient becomes profoundly neutropenic, with Escherichia coli and Klebsiella pneumoniae leading to bacteremia in only a moderate number of patients and other gram-negative bacilli rarely progressing to bacteremia despite colonization. Hence, the leading causes of bacteremia in the granulocytopenic patient are E. coli, K. pneumoniae and P. aeruginosa. Further investigations demonstrated that gram-negative bacilli were acquired from hands, food, and water, thus leading to approaches to infection prevention which included careful handwashing, low-microbial-content diet, and attention to water sources, including ice machines. Another basic approach to infection prevention has been to suppress gram-negative bacilli colonizing the alimentary canal with oral nonabsorbable antibiotics or, more recently and more effectively, with agents such as the fluoroquinolones which, unlike previous regimens, do not concurrently suppress the anaerobic flora, hence maintaining colonization resistance. The third basic approach to infection prevention is to improve the host defense factors, principally by a more rapid return of circulating granulocytes with the use of colony-stimulating factors such as granulocyte/macrophage colony-stimulating factor or granulocyte colony-stimulating factor. As to therapy, the fundamental approach with presumed gram-negative bacteremia is the prompt institution of empiric antibiotic therapy when fever first develops in the setting of granulocytopenia. There is a short "window of opportunity" after which no therapy will be effective. Combinations of antibiotics such as a beta-lactam and an aminoglycoside are used for multiple reasons: to afford coverage in the event the pathogen, proves resistant to one of the agents, to afford a synergistic activity thus improving and prolonging the serum bactericidal activity, and to reduce the development of resistance. However, patients can be divided into two risk groups: those with granulocytopenia and a regenerating bone marrow and those with an aplastic marrow who will have persistent, profound (< 100 microliters) granulocytopenia.(ABSTRACT TRUNCATED AT 400 WORDS)