If future trials of acute lung injury/acute respiratory distress syndrome are to be rigorous, informative, and successful, a number of key design issues need to be considered. First, appropriate sample size and entry criteria must be selected. The present definitions of acute lung injury/acute respiratory distress syndrome are arbitrary and select a broad, heterogeneous patient population in which treatment effects may often be small, requiring much larger sample sizes than those of previous trials. The alternative approach, selecting a subset of patients in whom a larger benefit is anticipated, is potentially hazardous because the subset selection criteria are unproven. Second, it must be ensured that the therapy is tested against current best methods of care. To ensure that a study is considered current at completion, investigators should anticipate that recent evidence at study commencement will be considered standard at study completion. Up-to-date evidence-based medicine should therefore be encouraged for all enrolled patients and, probably, protocolized in unblinded studies. Multiple novel therapies can also be tested, but care must be paid to the particular study design choice. Third, appropriate outcomes must be chosen. The traditional end point of 28-day mortality is too short and too crude to capture all relevant patient and societal outcomes. Thus, consideration of survival over a longer duration, coupled with assessment of quality of life, functional status, and morbidity, is essential. Fourth, the study must comply with new standards for the protection of human subjects. Protecting human subjects' rights and ensuring patient safety in subjects who are critically ill and who are rarely able to provide fully informed consent is a significant challenge. However, it is essential that new studies comply with required standards without becoming so burdensome that they cannot reasonably be completed.