Background: Prone position ventilation improves oxygenation and reduces the mortality of patients with severe acute respiratory distress syndrome (ARDS). However, there has been some debate on the relationship between the improvement of oxygenation and survival. Herein, we analyze whether the improvement of oxygenation would predict survival in patients receiving prone positioning for ARDS. Methods: We retrospectively reviewed adult patients receiving prone positioning for ARDS in medical intensive care unit (ICU) at Seoul National University Hospital between 2017 and 2019. Demographic variables, laboratory examinations, ventilator setting and clinical outcomes were collected for analysis. Arterial blood gases were analyzed before, during, and after prone positioning session. The primary outcomes were ICU and 28-day mortality. A multivariate Cox regression was used to identify independent predictors of survival. Results: A total of 63 patients receiving prone positioning were included, of whom 22 (34.9%) were ICU survivor. Male were 46 (73%) and median age was 68 years (IQR 59-74). Although there was no difference in PaO2/FiO2 (P/F) ratio prior to turning prone between survivors and non-survivors (92.5 ± 30.6 vs. 95.4 ± 31.7, p=0.730), subjects who survived had a higher P/F ratio after 8 hours of prone positioning (170.2 ± 69.5 vs. 110.0 ± 47.6, p=0.001). The area under the receiver operating characteristic (ROC) curve for delta-P/F ratio to predict ICU mortality was 0.863 (sensitivity 76.9% /specificity 85.7%) with an optimal cut-off value of an increase in P/F ratio of 31.6%. Prone responders were defined as an increase in P/F ratio of ≥31.6%. In the multivariate Cox regression analysis, prone responders (HR=0.236, 95% CI, 0.118-0.470, p<0.001) and SOFA score (HR=1.196, 95% CI, 1.070-1.336, p=0.001) were independent factors in predicting 28- day mortality. Conclusions: Improvement of oxygenation after 8hr of prone positioning may be good predictor for survival in the patients receiving prone positioning for ARDS.