OBJECTIVE: To evaluate the hypothesis that early-onset sepsis increases risk of death or neurodevelopmental impairment (NDI) among preterm infants; and that among infants without early-onset sepsis, prolonged early antibiotics alters risk of death/NDI. STUDY DESIGN: Retrospective cohort study of infants born at the Eunice Kennedy Shriver National Institute of Child Health and Human Development Neonatal Research Network centers (2006-2014) at 22-26 weeks of gestation and birth weight 401-1000 g. Early-onset sepsis defined as growth of a pathogen from blood or cerebrospinal fluid culture ≤72 hours after birth. Prolonged early antibiotics was defined as antibiotics initiated ≤72 hours and continued ≥5 days without culture-confirmed infection, necrotizing enterocolitis, or spontaneous perforation. Primary outcome was death before follow-up or NDI assessed at 18-26 months corrected age. Poisson regression was used to estimate adjusted relative risk (aRR) and CI for early-onset sepsis outcomes. A propensity score for receiving prolonged antibiotics was derived from early clinical factors and used to match infants (1:1) with and without prolonged antibiotic exposure. Log binomial models were used to estimate aRR for outcomes in matched infants. RESULTS: Among 6565 infants, those with early-onset sepsis had higher aRR (95% CI) for death/NDI compared with infants managed with prolonged antibiotics (1.18 [1.06-1.32]) and to infants without prolonged antibiotics (1.23 [1.10-1.37]). Propensity score matching was achieved for 4362 infants. No significant difference in death/NDI (1.04 [0.98-1.11]) was observed with or without prolonged antibiotics among the matched cohort. CONCLUSIONS: Early-onset sepsis was associated with increased risk of death/NDI among extremely preterm infants. Among matched infants without culture-confirmed infection, prolonged early antibiotic administration was not associated with death/NDI.