Infants with congenital diaphragmatic hernia (CDH) demonstrate a wide range of anatomic and physiologic abnormalities, making it difficult to compare the efficacy of new forms of therapy such as extracorporeal membrane oxygenation (ECMO) among institutions. This study was undertaken to determine whether any predictors of severity could be identified in the ECMO era. The charts of all patients with CDH treated at this institution since 1984, when ECMO became available. (n = 110), were reviewed. Infants were considered high risk and included in this study if they presented with respiratory distress within the first 6 hours of life (n = 94). In order to focus on predictors of pulmonary insufficiency, patients who died of nonpulmonary causes or had other significant congenital anomalies were excluded from this review, leaving 59 patients for analysis. All the infants during this period had intensive pharmacological and ventilatory support. When needed, ECMO was offered postoperatively from 1984 to 1987, and preoperatively from 1987 to the present. Forty-five of 59 had a best postductal PO2 (BPDPO2) greater than 100 mm Hg, and 41 of these responders survived (91%). Fourteen patients had a BPDPO2 less than 100 mm Hg and only one survived (7%) (P = .0001). Mean BPDPO2 between survivors with or without ECMO, and nonsurvivors were also significantly different (P = .001). To incorporate ventilatory information, an oxygenation/ventilation index was devised: [OVI = PO2/(mean airway pressure x respiratory rate) x 100]. Differences in OVI between these three groups were also significant. When analyzing the data by the method proposed by Bohn (PCO2 v VI), no correlation between ventilatory parameters and outcome was found.(ABSTRACT TRUNCATED AT 250 WORDS)