Objective: To determine if very preterm infants with O2 dependency at 28 days of life, but not at 36 weeks gestational age, are at increased risk at 18 months. Population: 217 infants born in a tertiary center at 24-28 weeks' gestation in 1987-1992, divided into 3 groups: neonatal comparison (C) group: O2 ≤ 28 days of life (n = 76); bronchopulmonary dysplasia (BPD) -1: O2 > 28 days but not at ≥ 36 weeks gestational age (n = 48); and BPD-2: O2 ≥ 36 weeks (n = 93). Outcome measures: Growth, persistent respiratory problems (asthma, tracheostomy, home O2), surgery, hospitalisations, and neurodevelopmental impairments. Results: No difference in weight, height or head circumference, incidence of total hospital admissions, admissions for respiratory problems or to the intensive care unit, persistent respiratory problems, cerebral palsy or sensory impairment. BPD-2 children had more hernia repairs (C: 9% vs BPD-1: 5% vs BPD-2: 18%, p <.01), had more admissions per child (C: 0.77 vs BPD-1: 0.79 vs BPD-2: 1.30, p <.01), a lower mean DQ (C: 96.2 ± 16.6 vs BPD- 1: 97.5 ± 13.4 vs BPD- 2: 90.1 ± 20.1, p <.05) and a higher incidence of DQ < 90 (C: 18% vs BPD-1: 19% vs BPD-2: 35%, p<.05). Conclusion: BPD-1 children are similar in all respect at 18 months to comparison children. BPD-2 children are similar to the other groups at 18 months in growth, general health, pulmonary morbidity and neurologic outcome but differ in having a higher risk of number of admissions per child, more hernia repairs, and more developmental delays. For very low gestational age infants, we support the use of O2 need at 36 weeks when defining BPD.