Abstract： Objective To investigate the predictive values of different critical scoring systems for mortality in patients with severe acute respiratory failure (ARF) supported by venovenous extracorporeal membrane oxygenation (VV-ECMO).Methods Forty-two patients with severe ARF supported by VV-ECMO were enrolled from November 2009 to July 2015.There were 25 males and 17 females.The mean age was (44 ± 18) years (rang 18-69 years).Acute Physiology and Chronic Health Evaluation (APACHE) Ⅱ],Ⅲ,Ⅳ,Simplified Acute Physiology Score Ⅱ (SAPS) Ⅱ,Sequential Organ Failure Assessment (SOFA),ECMO net,PRedicting dEath for SEvere ARDS on VVECMO (PRESERVE),and Respiratory ECMO Survival Prediction (RESP) scores were collected within 6 hours before VV-ECMO support.The patients were divided into the survivors group (n =17) and the nonsurvivors group (n =25) by survival at 180 d after receiving VV-ECMO.The patient clinical characteristics and aforementioned scoring systems were compared between groups.Scoring systems for predicting prognosis were assessed using the area under the receiver-operating characteristic (ROC) curve.The Kaplan-Meier method was used to draw the surviving curve,and the survival of the patients was analyzed by the Log-rank test.The risk factors were assessed for prognosis by multiple logistic regression analysis.Results (1) Positive end expiratory pressure (PEEP) 6 hours prior to VV-ECMO support in the survivors group [(9.7 ±5.0)cmH2O,(1 cmH2O =0.098 kPa)] was lower than that in the nonsurvivors group [(13.2 ± 5.4)cmH2 O,t =-2.134,P =0.039].VV-ECMO combination with continuous renal replacement therapy(CRRT) in the nonsurvivors group (32％) was used more than in the survivors group (6％,x2 =4.100,P =0.043).Duration of VV-ECMO support in the nonsurvivors group [(15 ± 13) d] was longer than that in the survivors group [(12 ± 11) d,t =-2.123,P=0.041].APACHE]Ⅱ,APACHE Ⅲ,APACHE Ⅳ,ECMO net,PRESERVE,and RESP scores in the survivors group were superior to the nonsurvivors group (all P ＜ 0.05).(2) The areas under the ROC curve of APACHE Ⅳ score for predicting death were largest (0.792 ± 0.076,95％ CI:0.643-0.940,P ＜ 0.05).The best cutoff point was 48 for APACHE Ⅳ score with a sensitivity of 92.0％,specificity of 64.7％,and overall accuracy of 81％.(3) Kaplan-Meier survival analysis showed that 180 d survival rate of the low APACHE Ⅳ score group was higher than the high APACHE Ⅳ score group (x2 =11.331,P ＜ 0.05).(4) Multiple logistic rcgression analysis showed that PEEP (OR =1.555,95％ CI:1.097-2.204,P ＜ 0.05),APACHE Ⅳ score (OR =1.152,95 ％ CI:1.021-1.301,P ＜ 0.05),and PRESERVE score (OR =4.984,95％ CI:1.531-16.227,P ＜ 0.05) were independent risk factors associated with mortality of patients supported by VV-ECMO.Conclusion The critical scoring systems proved to have good prognostic ability in predicting hospital mortality for severe ARF patients supported by VV-ECMO.Compared to other scoring systems,APACHE Ⅳ score system predicted more accurately,while specific scoring systems in predicting hospital mortality showed no advantage.