Infectious complications are an important cause of morbidity and mortality in cancer patients, especially those receiving chemotherapy. Furthermore, neutropenia, fever and infection limit the dose-intensity of antineoplastic chemotherapy in cancer patients. Fever without clinical signs of a localized infection is the commonest clinical presentation in neutropenic patients. Early empirical administration of broad-spectrum antibiotics at the onset of fever has become common practice, but the specific empirical regimen remains controversial. Guidance from therapeutic clinical trials is not straightforward, since it is difficult to compare trials, due to major differences and deficiencies in their design and analysis. Clinical trials fall into two categories: (i) explanatory trials which assess the hypothesis under ideal conditions, and (ii) pragmatic trials, which assess the regimen under the conditions of clinical practice. Methodological issues that are of crucial importance in the recognition of limits and value of the results of clinical trials in this field are discussed. The EORTC-IATCG has performed nine large therapeutic trials of empirical antibacterial and antifungal therapy in febrile, neutropenic patients with cancer. The results of trials, V, VIII, IX and XI are reviewed, and issues to be resolved in future trials are also considered.