There is a striking paradox in the literature supporting high-profile measures to reduce ventilator-associated pneumonia (VAP): many studies show significant reductions in VAP rates but almost none show any impact on patients' duration of mechanical ventilation, length of stay in the intensive care unit and hospital, or mortality. The paradox is largely attributable to lack of specificity in the VAP definition. The clinical and microbiological criteria for VAP capture a population of patients with an array of conditions that range from serious to benign. Many of the benign events are manifestations of bacterial colonization superimposed upon pulmonary edema, atelectasis, or other non-infectious processes. VAP prevention measures that work by decreasing bacterial colonization preferentially lower the frequency of these mislabelled, more benign events. In addition, misclassification obscures detection of an impact of prevention measures on bona fide pneumonias. Together, these effects create the possibility of the paradox where a prevention measure may have a large impact on VAP rates but minimal impact on patients' outcomes. The paradox makes changes in VAP rates alone an unreliable measure of whether VAP prevention measures are truly beneficial to patients and behooves us to measure their impact on patient outcomes before advocating their adoption.