Objective: We investigated whether the survival of in-hospital cardiac arrest victims could be extended by extracorporeal cardiopulmonary resuscitation (E-CPR) supported with extracorporeal membrane oxygenation (ECMO) compared to conventional CPR (C-CPR). Design: A retrospective, single-center, observational study. Setting: A tertiary care university hospital. Patients: We retrospectively analyzed a total of 479 adult patients with witnessed in-hospital cardiac arrest from 2003 to 2008 (81 in the E-CPR group and 398 in the C-CPR group). Measurements and Main Results: The primary endpoint was in-hospital survival. In a multivariate regression model, E-CPR (odds ratio [OR] 6.83, 95% confidence interval [CI] 2.45 to 19.07, p < .001) was significantly associated with survival. Propensity score matching of patients with CPR ��� 10 min was used to balance the baseline characteristics and CPR variables that could potentially affect prognosis. In the matched population (n = 98), the in-hospital survival rate of the E-CPR group was significantly higher than C-CPR group 30.6% and 10.2%, respectively (OR 3.88, 95% CI 1.28 to 11.74, p = .016) In addition, there was a significant difference in the six-month survival between these two groups (Hazard ratio 0.62, 95% CI 0.40 to 0.98, p = .012 by log-rank test). Conclusions: E-CPR was a significant predictor for survival of patients with witnessed in-hospital arrest. It showed a survival benefit over C-CPR in patients with CPR of more than 10 min.