Background: Meta-analyses suggest that Vit. A supplementation may reduce CLD and sepsis in ELBW infants. Our pilot studies supported use of a higher dose (5,000 IU given IM 3 times/wk for 4 wks) than that used in most neonatal trials. Purpose: To assess the effectiveness and safety of this regimen for infants 401-1000g BW who required oxygen or mechanical ventilation at 24h age. Methods: Control infants received a sham procedure to mask clinical staff and examiners. All infants had weekly physical exams for signs of toxicity; a sample (n=301) had serum analyses in a central lab with documented reliability. A sample size of 800 infants was planned to identify a ≥20% reduction (alpha error 0.05; power 80%) in CLD(oxygen dependence at 36 wks postmenstrual age) or death by 36 wks. Results: 405 Vit. A infants and 402 Controls with a similar mean BW(770, 769g) and GA (26.8, 26.7wks) were enrolled at 24-96 h age.Effectiveness: Death before discharge occurred in 17% of both groups. CLD or death by 36 wks--the primary outcome variable--occurred less often in Vit. A infants than Controls (55% vs. 62%; p=0.029; Relative Risk[RR] =0.89, 95%CI=0.80-0.99). Similar albeit not significant risk reductions were noted for sepsis (RR = 0.90; 95%CI=0.76-1.07), periventricular leukomalacia (PVL) (RR=0.74; 95%CI=0.44-1.25), and the combined outcome of death, Grade 3/4 intracranial hemorrhage, or PVL (RR=0.92; 95% CI=0.76-1.12).Safety: Questionable or definite increased fontanelle tension--considered the most sensitive sign of Vit. A toxicity--was found in 15% of Vit. A infants and 18% of Controls on any exam (p=0.25). Vit. A toxicity was suspected from physical findings in 1.0% of Vit. A infants and 0.8% of Controls (p=0.72). At study day 28, low serum retinol values(<20μg/dl, a value <5 percentile in U.S. children) were observed in a sizable proportion of Vit. A infants (24%) albeit a lower percentage than in Controls (55%) (p<0.001). A high retinol dose response suggesting low or absent stores of vit. A (>10% increase 3h after 2000 IU/kg IM) was observed in 22% of Vit. A infants and 45% of Controls Conclusion: Given the significant though modest reduction in CLD, the high safety and low cost of the intervention, and the lack of evidence supporting conventional vitamin A regimens, our findings support routine clinical use of the vitamin A regimen tested for infants <1000g BW requiring respiratory support at 24h age. Because even supplemented infants may be vitamin A deficient, higher doses deserve future study.