Progress has been made in the development and validation of rules that attempt ito predict a low risk (<10%) of severe infection or clinical complications in patients with cancer, fever, and neutropenia. It is uncertain, however, which model is optimal, with respect to test characteristics, applicability, and interinstitutional reliability, and prospective model validation in a multicenter context among outpatients has not been performed. Clinical criteria, such as comorbidities and performance status, remain critical in the risk-assessment process and probably are used by most physicians caring for patients with cancer who are febrile and neutropenic. Clinical prediction rules might be improved in the future by including measurements of inflammatory markers, such as procalcitonin. Reliable prediction of the risk of medical complications may be relevant for decisions regarding parenteral versus oral antimicrobial therapy, but it is definitely needed for decisions regarding site of care. Site-of-care decisions require thorough assessment not only of medical criteria, but also of psychosocial and organizational and/or logistic criteria. If the appropriate infrastructure to provide follow-up is available, home-based therapy with oral (or parenteral) antibiotics is an acceptable option in the care of patients with cancer who have intercurrent febrile neutropenia and a predicted low risk for medical complications.