Determinations of the type and setting of empirical therapy for immunocompromised patients with fever are complicated by the characteristics of the underlying illness and the effects of treatments already received, as well as by changing microbiological patterns and trends in drug resistance at national and institutional levels. Several systems have been proposed to distinguish patients who could benefit from outpatient antibiotic therapy from patients who require hospitalization. Practical considerations may decide whether the necessary monitoring during the period of neutropenia can be achieved.