In view of these very encouraging results, should piperacillin-tazobactam become the standard therapy for all patients with VAP? The authors adequately caution against this view. In addition to the presence of risk factors that are specific to the underlying disease of patients, it has become clear that there are three important determinants of the etiology of VAP that can be used to guide the selection of initial therapy, as follows: the timing of onset of pneumonia relative to ICU (or preferably hospital) admission; the prior administration of antibiotics; and unit-specific local epidemiologic factors. The distinction between early pneumonia (ie, after < 5 to 7 days in the ICU) and late-onset pneumonia is a major determinant of the etiology, provided that the patient has not spent some time in the hospital before ICU admission or has had a recent hospitalization. In early-onset cases, the clinician can be fairly confident that “normal flora” (ie, Streptococcus pneumoniae and other streptococci, H influenzae, S aureus, and possibly anaerobes in selected circumstances) are involved, and there is no need for the administration of an antipseudomonal, broad-spectrum penicillin combined with a β-lactamase inhibitor or of ciprofloxacin. Therapy with a first-generation cephalosporin or a combination of ampicillin with a β-lactamase inhibitor will suffice. However, if the patient has received antibiotics for some time, especially broad-spectrum ones,12 then the flora colonizing and infecting the respiratory tract is much more likely to have been modified to include “hospital organisms,” such as Klebsiella, Enterobacter, and particularly P aeruginosa, all of which are likely to be resistant to any antibiotics that had been administered previously. The worst-case scenario is for the patient to have received several extended courses of antibiotics, and, when presenting with pneumonia occurring late after hospital/ICU admission, to be exposed to an environment in which resistant organisms such as Acinetobacter, Pseudomonas, methicillin-resistant S aureus, or vancomycin-resistant Enterococcus are endemic. In this instance, it is very difficult to predict the microorganisms that may be involved and their susceptibility profile, so that a three-agent combination empiric therapy with a β-lactam, an aminoglycoside, and a glycopeptide is commonly prescribed pending culture results, unless a more targeted approach can be guided by the direct examination of reliable respiratory tract specimens showing a single population. In such settings, piperacillin-tazobactam or any other single drug is unlikely to cover adequately all microorganisms involved.