Although the results of the current study are interesting, there are some shortcomings to the study design, as the authors acknowledge. First, this was not a randomized trial testing different ventilation strategies for acute lung injury; therefore, those patients receiving higher tidal volumes may have been treated differently from those receiving lower tidal volumes in ways that were not measured in this study. For instance, as the authors point out, the use of many treatment strategies found in recent trials to decrease mortality in sepsis (such as activated protein C, tight control of glucose, and corticosteroids for relative adrenal insufficiency) was not recorded. Secondly, the duration of time that patients with lung injury were exposed to tidal volumes > 7.4 mL/kg of PBW was not available, nor was the plateau pressure reported in these patients. Thirdly, the data were recorded only once per day, introducing the possibility that some patients had significant fluctuations in their ventilatory strategy not measured by this analysis. Finally, mean tidal volume did not differ significantly between survivors and nonsurvivors. As the authors point out, this finding may be due to the fact that mean values by their nature do not describe the degree of variation around the mean but rather reflect a simple average of all the values in the sample, thus tending to obscure variation. While this explanation may be correct, the lack of significant difference between the mean tidal volume in survivors and nonsurvivors does raise some concern about the conclusions of the study. Despite these limitations, the inclusiveness and size of the study population and the strength of the multivariable analysis suggest that the findings of this study should be considered seriously.