Currently, steroid is not advocated in acute respiratory distress syndrome (ARDS). The effects of steroid on the clinical outcome of refractory severe ARDS remain uncertain. We investigated whether steroid has impacts on mortality of ARS treated with extracorporeal membrane oxygenation (ECMO). Methods: We retrospectively analyzed 441 patients with ARDS who underwent ECMO therapy from 11 hospitals in South Korea from January 2012 to December 2015. Patients were divided into steroid (n=60) and non-steroid groups (n=381), and in-hospital mortality was compared in two groups after propensity score matching. We used cox regression to examine factors associated with in-hospital mortality. Results: Overall, in-hospital mortality was significantly higher in steroid group (73.3% vs 56.4%, p=0.013). The steroid group had significantly higher age (59.7 ±14.8 vs 54.9±16.1, p=0.031), lower respiratory ECMO survival prediction score (-1.3±3.8 vs 0.1±3.9, p=0.010), and longer mechanical ventilation time before ECMO initiation ( 7 days: 18.3% vs 14.4% p=0.017). After propensity score matching, in-hospital mortality did not differ between two groups (70% vs 65%, p=0.323). Otherwise, subgroup analysis showed that steroid was independently associated with increased in-hospital mortality in patients who were successfully weaned off from ECMO after adjusting baseline variable such as age, initial sequential organ failure assessment score, prone positioning before ECMO initiation, and pre PaO2/FiO2 ratio (HZ 2.33, 95% CI 1.23-4.42, p=0.010). Steroid is associated with deterioration after successful weaning off from ECMO.