In 1991, Schneider and Voerman69 were the first investigators to suggest that “physiologic and not pharmacologic doses of glucocorticoids [be administered] in the course of septic shock”. These authors demonstrated reversal of shock in 3 of 8 patients given “100 mg of hydrocortisone IV followed by 100 mg every 8 h with dose tapering with improvement.” The use of extended course, stress-dose corticosteroids has been evaluated in 10 RCTs in critically ill patients with sepsis, septic shock, and ARDS (Table 2).30–33,70–75 Overall, this dosing strategy has been reported to be associated with a significant reduction in 28 day all-cause mortality, more rapid weaning of vasopressor agents (septic shock), a reduction in ICU length of stay, and an increase in ventilator-free days (ARDS).28,29,34,76 While many clinicians may be reluctant to prescribe corticosteroids based on the “negative” results of the recent CORTICUS study,75 it should be appreciated that this study has a number of serious limitations. Most notably, the lack of clinical equipoise resulted in a significant selection bias (patients least likely to benefit from corticosteroids were enrolled in the study).77 The ARDSnet Late Steroid Rescue Study investigated the role of a supra-physiologic dose of methylprednisolone (approximately 700 mg hydrocortisone equ/day) in patients with severe persistent ARDS after 7 days of ventilatory support.78 While the design and outcome of this study has been much debated,76,79 there was a significant increase in ventilator-free days (alive and off mechanical ventilation) in the corticosteroid group, although overall mortality was not improved. It is, however, important to recognize that both the Late Steroid Rescue Study and the study by Meduri et al33 enrolled patients before lung protective ventilatory strategies were widely adopted. Since high tidal volume ventilation has been documented to increase the production of proinflammatory mediators,80,81 it is possible that the benefit of corticosteroids may be limited to patients not receiving a protective ventilatory strategy. Furthermore, while the data in Table 2 and previously published systematic reviews28,29,76 suggest that corticosteroids may be beneficial in critically ill patients with severe sepsis, septic shock, and ARDS, the studies included in these analyses are limited by sample size and methodologic issues (eg, use of etomidate), and therefore, the risk-to-benefit ratio of stress-doses of corticosteroids should be assessed in each individual patient. However, based on current evidence, it would be reasonable to initiate treatment with stress-doses of corticosteroids in patients with vasopressor-dependent septic shock (norepinephrine or equ requirement > 0.05 to 0.1 μg/kg/min) within 12 h of presentation and in patients with persistently severe ARDS who have failed to improve after 48 h of supportive care and a lung protective strategy (Table 3). A patient who requires only a few hours of low-dose vasopressor support while being fluid resuscitated is, however, unlikely to benefit from corticosteroids. While previous studies suggested that the decision to treat patients with septic shock should be based on the results of an ACTH stimulation test, the limitations of this test in diagnosing CIRCI and the benefit of corticosteroids in both responders and nonresponders suggest that this test should not be used to select patients likely to derive benefit from corticosteroids.34 It is, however, unclear at this time whether dynamic testing of the HPA axis should be performed once patients have recovered from their acute illness and completed a course of corticosteroids to determine the normalization of the HPA axis. Furthermore, additional studies are required to determine whether the treatment of CIRCI improves long term mortality. It is possible that CIRCI is an epiphenomenon and a marker of illness severity, and that treatment with corticosteroids may alter the expression of inflammatory mediators and the severity of the inflammatory response without altering ultimate patient outcome.