Febrile neutropenia is still associated with a high mortality rate, making timely and efficient empirical antibiotic therapy absolutely vital. For these reasons, evidence-based guidelines are urgently needed. The guidelines published so far are mainly based on clinical experience and selective citation. This review summarises studies and meta-analyses concerning empirical antibiotic therapy in high-risk neutropenic patients: (1) No benefit results from the addition of an aminoglycoside to the initial empirical therapy. On the contrary, patients who received an aminoglycoside had a significantly higher rate of adverse events, especially nephrotoxicity. (2) The empirical addition of a glycopeptide after 3-4 days of persistent fever was evaluated in two randomised controlled trials. Combined analysis demonstrates that in clinically stable patients without resistant or skin/soft tissue infections, the use of a glycopeptide can be delayed for another 3-4 days. (3) The choice of drugs for monotherapy is currently being evaluated; preliminary results demonstrate that ceftazidime has a significantly inferior response rate (without modification) to other evaluated antibiotics. In conclusion, guidelines should be based on the systematic evaluation of all relevant clinical trials. The analysis of the existing data leads to the recommendation of monotherapy, without aminoglycoside, using piperacillin-tazobactam, cefepime, meropenem or imipenem-cilastin, any of which may be continued for up to 7 days in persistently febrile, clinically stable patients without skin/soft tissue infections. The choice of drug as standard first-line therapy should depend on drug costs, local resistance rates and the potential for resistance induction.