Use of extracorporeal membrane oxygenation for treatment of respiratory failure caused by sepsis is controversial because of concerns over survival benefit and hemorrhage-related complications. To evaluate the impact of the primary diagnosis of sepsis on outcome, we reviewed data from 6853 neonates in the Extracorporeal Life Support Organization Registry and defined two groups: group 1 (n = 1060), all patients undergoing extracorporeal membrane oxygenation with a primary diagnosis of sepsis; group 2 (n = 5793), those with any other primary diagnosis. A multivariate logistic regression analysis that considered 15 variables present before extracorporeal membrane oxygenation (including age, sex, birth weight, prior cardiopulmonary arrest, arterial blood gas results, and ventilator settings) was used to compare outcomes between groups. Survival was not different between the two groups (77%, group 1; 82%, group 2; p = 0.2480), although lung recovery was less frequent in the patients with sepsis (p = 0.0185). Group 1 had a higher incidence of complications including seizures (odds ratio 1.446, p = 0.0346), cerebral infarct or hemorrhage (2.310, p = 0.0001), need for dialysis (1.478, p = 0.0131), hypernatremia (2.089, p = 0.0019), hyperbilirubinemia (2.423, p = 0.0001), and dobutamine use (1.918, p = 0.0001). Neonates with sepsis are more likely to have neurologic, renal, and metabolic complications from extracorporeal membrane oxygenation but may still achieve a survival benefit equivalent to those without sepsis. From these data, extracorporeal membrane oxygenation should not be withheld from neonates solely on the basis of sepsis. Rather, management strategies should focus on limiting the incidence or severity of the common complications.