Abstract. Objective: Transport of infants born at the threshold of viability (<27 weeks) may negatively impact outcomes. Our neonatal intensive care unit (NICU) is an all-referral unit, and therefore all patients have been transported. This study examined survival rates and the factors that influenced survival in this specific population of extremely premature infants. Study design: The study setting was at an all-referral tertiary care NICU within a children's hospital. Data were collected from December 2004 to August 2008, and included 227 patients. Results: Overall, the survival-to-discharge rate for these infants was 78%. The primary causes of death were sepsis (41%), necrotizing enterocolitis (NEC) (27%), respiratory failure (18%), and severe intraventricular hemorrhage (IVH) (10%). By logistic discrete time hazard analyses, patients receiving dopamine (p < 0.01) or insulin (p < 0.01), with NEC (p < 0.05), or of gestational age 22, 23 or 24 weeks (p < 0.01) were more likely to die; while those born in the same county as the NICU (p < 0.01), receiving patent ductus arteriosis (PDA) treatment (p < 0.01) or peripheral intravenous central catheters (p < 0.01) were more likely to survive. Within the group of survivors the rate of cerebral palsy was 9% and the mean scores on the Bayley-III at 18 months corrected age were 95 ± 13, 90 ± 15, and 92 ± 14 in the cognitive, communication, and motor scales respectively. Conclusion: Infection, NEC, and respiratory failure were the primary causes of death among extremely premature infants in an all-referral NICU. New approaches to preventing infection, NEC, and bronchopulmonary dysplasia are needed for this population.