Background: We evaluated retrospectively the early and midterm results of using veno-arterial extracorporeal membrane oxygenation (VA-ECMO) support in patients undergoing valvular surgery. Methods: A total of 87 patients undergoing valvular surgery received VA-ECMO due to refractory postcardiotomy cardiogenic shock (PCS), who were eligible for inclusion were enrolled in this study. Preoperative, perioperative, and postoperative variables were assessed and analyzed for possible associations with mortality in hospital and after discharge. Results: The mean age, additive EuroSCORE, and left ventricular ejection fraction (LVEF) for all patients was 65 +/- 7 years, 6.1 +/- 1.9 points, and 46% +/- 12%, respectively. The mean duration of VA-ECMO support was 61 +/- 37 hours. Intra-aortic balloon pumps (IABP) were implanted in 47.1% of patients. Weaning from VA-ECMO was successful in 59% of patients, and 49% were discharged. Multivariate analysis revealed that being >65 years old (odds ratio [OR], 2.75), receiving postoperative renal replacement treatment (OR, 2.47), having a peak lactate level >= 12 mmol L-1 (OR, 2.18), and receiving VA-ECMO for >60 hours (OR, 3.2) were independent predictors of in-hospital mortality. IABP support (OR, 0.46) was protective. In addition, persistent heart failure with an LVEF <40% was an independent predictor of mortality after discharge. Conclusions: VA-ECMO is an acceptable technique for the treatment of PCS in patients undergoing valvular surgery, who would otherwise die. It is justified by the good long-term outcomes of hospital survivors, but the use of VA-ECMO must be decided on an individual risk profile basis because of high morbidity and mortality rates.