Dr. Savel and colleagues correctly argue that recommendations for practice may not rely exclusively on the results of a single trial. For this reason, my recommendation for the routine use of low-dose corticosteroids in combination with secondary prophylaxis in patients with acute lung injury/ARDS was based on strong experimental and translational data and the consistent findings of five randomized controlled trials.1One may argue that these five trials were relatively small in size and that not all of the studies showed a significant survival benefit. However, a metaanalysis integrating the outcome data of these trials shows that prolonged glucocorticoid treatment initiated before day 14 of ARDS is associated with a significant reduction in mortality (84 of 252 patients [33%] vs 108 of 216 patients [50%]; relative risk, 0.71; 95% confidence interval, 0.50 to 0.87; p = 0.001). What should clinicians do when no large randomized controlled trial is available and a metaanalysis of small sized trials shows significant reduction in mortality that can also be supported by biological plausibility? It is this author’s opinion that while waiting for a large-scale trial to investigate mortality as a primary end point, patients with acute lung injury/ARDS should be treated with low-dose corticosteroids because this treatment undoubtedly reduces morbidity and the duration of mechanical ventilation. Additionally, potential glucocorticoid side effects can be prevented by strict surveillance for superinfection and blood glucose control, further increasing the treatment benefit/risk ratio. Regarding the internal validity of the trial of Meduri and colleagues,2 having a treatment allocation based on 2:1 principle is not a problem, assuming that placebo-treated patients will behave roughly similarly. That more patients in the placebo group were in shock was due to chance, and per-protocol analysis failed to reproduce this imbalance in the proportion of vasopressor-dependent patients. In addition, this would suggest that the positive effects of methylprednisolone in this population were unlikely entirely related to shock reversal. As far as mechanical ventilation is concerned, the proportion of patients receiving mechanical ventilation at day 28 was affected by the introduction of open-label methylprednisolone in those failing to improve (8% vs 36%, p = 0.002). Most importantly, the difference in mechanical ventilation-free days at day 28 (16.5 ± 10.1 days vs 8.7 ± 10.2 days, p = 0.001) underscores the significant impact of treatment on disease resolution and supports my argument for the routine use of this protocol in conjunction with measures demonstrated to reduce morbidity associated with glucocorticoids.