The objective of this retrospective analysis was to study the possibility of a treatment strategy for febrile, granulocytopenic patients based on relative infective risks. Data from 441 adult febrile, granulocytopenic patients treated initially with antibacterial monotherapy were reassessed in terms of infection risk and outcome. Two different risk-groups were identified, namely those without and those with a clinical focus of infection (e.g. the respiratory tract and skin and soft tissue). Patients with a focus took longer to respond (4-8 days to defervesce) than those with unexplained fever, urinary tract infection or bacteraemia (2-3 days to defervesce). Patients with respiratory tract and skin and soft tissue infections received additional antibiotics more frequently (greater than 15% versus less than 10%) and were treated for a longer period (50% greater than or equal to 9 versus less than or equal to 8 days). More often the bone marrow had recovered at the time of response (48% versus 65%). These data suggest that a differentiated approach in the selection of initial antibiotics may be feasible, based upon continuous assessment of the patient's clinical condition. Antibiotic therapy may be postponed in patients without a focus of infection who show slowly increasing temperature. Patients with upper respiratory tract and skin and soft tissue infections may benefit from initial addition of penicillin and a glycopeptide, respectively. In cases of lower respiratory infections, antifungal therapy should be considered at an early stage.