Infants with congenital diaphragmatic hernia (CDH) show a wide range of anatomic and physiological abnormalities, making it difficult to compare the efficacy of management protocols between institutions. The purpose of this study was twofold: (1) to analyze the results of treatment of CDH in a large tertiary care pediatric center using conventional mechanical ventilation (CMV) with extracorporeal membrane oxygenation (ECMO) as rescue therapy, and (2) to compare these results with those of a parallel study by a similar large urban center that used high-frequency oscillating ventilation (HFOV) as rescue therapy without ECMO. All patients who had CDH diagnosed within the first 12 hours of life and were referred for treatment before repair (between 1981 and 1994) were included in the analysis (n = 196). CMV was used initially in all patients, with conversion to ECMO for refractory hypoxemia or hypercapnea. Between 1981 and 1984, ECMO was not available. Between 1984 and 1987, ECMO was offered postoperatively. Between 1987 and 1991, ECMO was offered preoperatively. In all three groups, aggressive hyperventilation and alkalosis was the norm. Since 1991, permissive hypercapnia has been used. HFOV was used in three patients as stand-alone therapy with one survivor. Twenty patients died without repair: Ten had other lethal anomalies, eight died before ECMO could be instituted, and two died of ECMO-related complications. Overall, 104 patients (53%) survived and 92 (47%) died. Ninety-eight patients (50%) received ECMO, and 43 (44%) survived. Survivors had significantly higher 1- and 5-minute Apgar scores and higher postductal Po2s than did nonsurvivors. Associated anomalies were present in 39%, who had a significantly lower survival than those with isolated CDH. Antenatal diagnosis and side of the defect had no impact on outcome. Survival was not improved with the institution of ECMO or delayed repair but rose significantly to 69% (84% with isolated CDH, P = .007) with the introduction of permissive hypercapnea. Autopsy results from nonsurvivors showed other lethal anomalies and significant barotrauma as the primary causes of death. Comparisons between the Boston and Toronto series showed similar patient demographics and no significant differences in survival in any time period. The two series differed in the number of associated anomalies, their impact on survival, and in the prognosis of right-sided CDH. From the individual and combined analyses the authors concluded: (1) CMV with ECMO as rescue produced an overall survival in CDH patients equivalent to CMV with HFOV in a parallel series, (2) neither HFOV nor ECMO has significantly improved outcome in CDH patients, (3) institution of permissive hypercapnia has resulted in a significant increase in survival, and (4) the leading causes of death in CDH patients appear to be associated anomalies and pulmonary hypoplasia, which are currently untreatable. Barotrauma, which may contribute in up to 25% of deaths in CDH patients is avoidable.