BACKGROUND: A National Institute of Child Health and Human Development Neonatal Research Network randomized trial showed that vitamin A supplementation reduced bronchopulmonary dysplasia (O2 at 36 weeks' postmenstrual age) or death in extremely low birth weight (ELBW) neonates (relative risk [RR]: 0.89). As with postnatal steroids or other interventions, it is important to ensure that there are no longer-term adverse effects that outweigh neonatal benefits. PRIMARY OBJECTIVE: To determine if vitamin A supplementation in ELBW infants during the first month after birth affects survival without neurodevelopmental impairment at a corrected age of 18 to 22 months. DESIGN/METHODS: Infants enrolled in the National Institute of Child Health and Human Development vitamin A trial were evaluated at 18 to 22 months by carefully standardized assessments: Bayley Mental Index (MDI) and Psychomotor Index (PDI), visual and hearing screens, and physical examination for cerebral palsy (CP). The medical history was also obtained. Neurodevelopmental impairment (NDI) was predefined as > or =1 of MDI <70, PDI <70, CP, blind in both eyes, or hearing aids in both ears. RESULTS: Of 807 enrolled infants, 133 died before and 16 died after discharge. Five hundred seventy-nine (88%) of the 658 remaining infants were followed up. The primary outcome of NDI or death could be determined for 687 of 807 randomized infants (85%). Baseline characteristics and predischarge and postdischarge mortality were comparable in both study groups. NDI or death by 18 to 22 months occurred in 190 of 345 (55%) infants in the vitamin A group and in 204 of 342 (60%) of the control group (RR: 0.94; 95% confidence interval: 0.80-1.07). RRs for low MDI, low PDI, and CP were also <1.0. We found no evidence that neonatal vitamin A supplementation reduces hospitalizations or pulmonary problems after discharge. CONCLUSION: Vitamin A supplementation for ELBW infants reduces bronchopulmonary dysplasia without increasing mortality or neurodevelopmental impairment at 18 to 22 months. However, this study was not powered to evaluate small magnitudes of change in long-term outcomes.