To assess the validity of the currently accepted lower weight limit of 2 kg for treatment of neonates with extracorporeal membrane oxygenation (ECMO), we reviewed the outcome of lower birth weight (2.0 to 2.5 kg, n = 29) and higher birth weight (n = 235) for infants treated with venoarterial ECMO at our institution from 1984 through 1990. Newborn infants with congenital diaphragmatic hernia were not included. The mortality rate was significantly greater after venoarterial ECMO in lower than in higher birth weight infants (relative risk 3.45; confidence interval = (1.68, 5.79)). For infants with the diagnosis of respiratory distress syndrome, the mortality rate was 56% (5/9) for lower and 8% (2/25) for higher birth weight infants (p less than 0.01). The most frequent cause of death in lower birth weight infants was intracranial hemorrhage (7/10 deaths). The overall incidence of any neuroimaging abnormality was significantly greater for lower birth weight infants (p = 0.044), primarily because of the higher incidence of major intracranial hemorrhage. Finally, the risk of developmental delay (development quotient less than 70 at 1 to 2 years of age) among survivors available for follow-up was significantly higher among the lower than the higher birth weight infants. These outcome data suggest that further reduction of the current lower weight limit for ECMO should not become standard without prospective research or technologic advances.