Bloodstream infections (BSI) are a significant cause of morbidity and mortality in onco-hematologic patients. The Gram-negative bacteria were the main responsible for the febrile neutropenia in the sixties; their impact declined due to the use of fluoroquinolone prophylaxis. This situation was followed by the gradual emergence of Gram-positive bacteria also following the increased use of intravascular devices and the introduction of new chemotherapeutic strategies. In the last decade, the Gram-negative etiology is raising again because of the emergence of resistant strains that make questionable the usefulness of current strategies for prophylaxis and empirical treatment. Gram-negative BSI attributable mortality is relevant, and the appropriate empirical treatment significantly improves the prognosis; on the other hand the adequate delayed treatment of Gram-positive BSI does not seem to have a high impact on survival. The clinician has to be aware of the epidemiology of his institution and colonizations of his patients to choose the most appropriate empiric therapy. In a setting of high endemicity of multidrug-resistant infections also the choice of targeted therapy can be a challenge, often requiring strategies based on off-label prescriptions and low grade evidence. In this review, we summarize the current evidence for the best targeted therapies for difficult to treat bacteria BSIs and future perspectives in this topic. We also provide a flow chart for a rational approach to the empirical treatment of febrile neutropenia in a multidrug resistant, high prevalence setting.