Although antibiotic therapy is the mainstay of therapy for gram-negative bacillary bacteremia, the amelioration of the underlying conditions, the correction of predisposing factors, the drainage of abscesses, the removal of infected foreign bodies, and adequate supportive care are also of paramount importance for curing the infection and should not be neglected. Beginning in the late 1960s, most of the clinical work on gram-negative infections has focused on the evaluation of new antibiotics. Numerous studies have shown that early, appropriate antibiotic treatment of gram-negative bacteremia significantly improved patients' outcomes and prevented the development of septic shock. Prescribing standard doses of antibiotics does not necessarily mean that therapeutic levels will be reached in all patients, and relapses of infections or breakthrough bacteremias can occur in patients with subinhibitory serum levels of antibiotics. The monitoring of serum concentrations of antibiotic is therefore recommended in critically ill septic patients. Whereas initial studies on the antibiotic treatment of gram-negative bacteremia were carried out in nonneutropenic patients, more recent clinical investigations have been performed almost exclusively in cancer patients with neutropenia. Studies conducted in the 1970s and 1980s among these patients have shown the following: (1) early empirical therapy reduced the mortality of gram-negative bacteremia; (2) therapy with a combination of two antibiotics, be it an extended spectrum penicillin plus an aminoglycoside or a third-generation cephalosporin, has significantly improved patients' outcomes; and (3) triple-drug combinations (i.e., a penicillin plus a cephalosporin plus an aminoglycoside) are not superior to combinations of beta-lactams and aminoglycosides. For the treatment of gram-negative bacteremia, clinicians today have a choice between well-established antibiotic combinations and broad-spectrum single-agent therapy with third-generation cephalosporins or carbapenem antibiotics. Although recent studies suggested that monotherapy could be as effective as combination therapy for the empirical treatment of fever in the neutropenic host, no definitive study has so far unquestionably demonstrated the equivalence of these treatments in patients with gram-negative bacteremias, especially those caused by P. aeruginosa, or in patients with adverse prognostic conditions, such as persistent and profound granulocytopenia. This literature should however be reviewed with great caution. Indeed, only a minority of studies have included a sufficient number of patients to confidently assess the impact of therapy on patients' outcomes. Obviously, small studies can have a significant risk of type II errors, that is, making false-negative conclusions.(ABSTRACT TRUNCATED AT 400 WORDS)