The PneumA investigators, to use their own words, sought recurrent infections with “extreme vigilance.”7 To classify a patient as having recurrent pneumonia, the investigators required positive quantitative cultures from bronchoscopic specimens. The triggers for bronchoscopy were extremely broad. They included new fever, purulent secretions, new or progressive pulmonary infiltrates, hemodynamic instability requiring initiating or increasing vasopressors, deterioration of the Pao2:Fio2 ratio by > 30%, and/or any change in antibiotic therapy, regardless of cause. Furthermore, although clinicians were initially blinded to patients’ treatment assignments, they were unblinded on day 8. So by definition, all possible relapses were sought by clinicians aware of patients’ antibiotic exposures. It is not difficult to imagine that some investigators might have been more prone to test for recurrent infection in patients who received short-course therapy, particularly during the second week postrandomization, when they knew that patients treated with short-course therapy were no longer being treated, while those in the long-course group remained on antibiotics. Furthermore, we know from other studies that endotracheal tubes are highly prone to colonization, that about 40% of patients with positive quantitative cultures for Pseudomonas have no clinical manifestations of infection, that sterilizing the endotracheal tube and respiratory tract is very difficult even with prolonged antibiotic courses, and that antibiotic therapy may paradoxically increase Pseudomonas colonization rates.2,9,10 Autopsy series have taught us that both BAL and protected specimen brush quantitative cultures have false-positives rates of about 30% and 40%, respectively.11-13 These observations make it highly likely that a significant fraction of the recurrent pneumonias in the PneumA study, and by extension in this meta-analysis, were instances of colonization rather than invasive infection. Comparable net outcomes for patients treated with short-course vs long-course therapy provide further evidence that the ostensibly higher relapse rate with short-course therapy may have been driven more by colonization than recurrent invasive infections.