Extracorporeal membrane oxygenation (ECMO) is used for severe neonatal and pediatric respiratory failure that fails to respond to maximal therapy, including but not limited to high frequency oscillatory ventilation (HFOV), surfactant replacement, inhaled nitric oxide (NO) and prone positioning of the patient. This report reviews our decade of experience with ECMO and describes various factors that affect the outcome of neonatal and pediatric ECMO for respiratory failure (RF). General clinical and technical concepts regarding ECMO use are also reviewed. We retrospectively evaluated 84 patients, 70 neonatal and 14 pediatric, who were treated with ECMO in our Childrens Hospital and showed that the outcome of these patients depends on their underlying pulmonary pathology and the degree of its clinical reversibility. We divided our neonatal and pediatric RF patients into 3 groups: patients with high disease reversibility (Group N1 for neonates and group P1 for pediatric patients), low disease reversibility (groups N2 and P2) and no predicted disease reversibility (groups N3 and P3). The characteristics of the patients and their outcome in these 3 categories are described. Overall, our neonatal ECMO had a better outcome than pediatric ECMO with survival to discharge rates of 76% and 29%, respectively. Patients with neonatal RF without surgically treatable problems in their lungs or ribcages (group N1) had a 92% survival to discharge rate, while neonates with congenital diaphragmatic hernia (group N2) have a survival rate of only 53%. Achieving sustained negative fluid balance during neonatal ECMO of patients in group N1 is an excellent predictor of ECMO success. In pediatric RF the noninfectious etiology for ARDS had a better prognosis (75%) than infectious etiology had (10%). We support the concept that patients who have received more than 7 days of mechanical ventilation for acute respiratory distress syndrome (ARDS) should not be considered for ECMO as their predicted outcome is extremely grave.