SummaryIn spite of considerable progress in clinical care and supportive measures, infection remains by far the principal cause of morbidity and mortality in febrile neutropenic patients, especially following intensive myelosuppressive therapy for haematological mali-gnancies. The concept of empirical therapy, dating from the early 1970s and referring to the use of broad- spectrum antibiotics without or before definite proof of infection, was of paramount importance for the eli-mination of early infectious deaths, mainly due to Gram-negative bacteria, and has dramatically reduced mortality figures by infection to about 6%. However, over the last two decades, consecutive trials have witnessed marked shifts, both in the spectrum of of-fending pathogens, bacterial as well as non-bacterial, and in the clinical presentations.Although most clinical studies completed by in-ternational groups have advocated the use of combi-nation therapy on theoretical grounds, including a beta-lactam plus an aminoglycoside or double beta- lactams, no single recent trial comparing monotherapy with the newer extended-spectrum compounds versus combination therapy could demonstrate relevant differences in outcome. Contrary, combinations were hampered by associated toxicity. We currently accept monotherapy with agents such as carbapenems and third-generation cephalosporins (ceftazidime) as standard of empirical care in neutropenic patients while the results of ongoing trials with ‘fourth-generation’ cephalosporins and fluoroquinolones are awaited with impatience. However, the universally adopted and indiscriminate implementation of broadspec- trum therapy has undoubtedly taken a toll of current anti-infective strategies, including the emergence of highly resistant isolates and the disastrous shift towards invasive mould and yeast infections. Consequently, empirical therapy has evolved towards a planned-progressive modification in persistent neutropenic febricity, especially aiming at early antifungal coverage. However, randomised data on second- and third-line modifications of empirical regimens are sparse; moreover, significant divergence exists between European and North-American centres. The exact place and need for glycopeptide antibiotics re-mains another controversial issue, certainly, in view of the alarming isolation of resistant enterococci.The improved identification of patient subgroups by the development of accurate risk assessment models based on prognostic factors, such as age, underlying disorder, duration of neutropenia, intensity of treatment and offending pathogen, should result in a more rational application of antimicrobial agents and may pave the way to a patient-tailored (rather than a planned-progressive) and modifiable anti-infective approach with respect to hospitalisation and need for extended-spectrum empirical therapy.