The purpose of this trial was to test the hypothesis that vitamin A adequacy may decrease the incidence of bronchopulmonary dysplasia (BPD) in very-low-birth-weight infants. Serum vitamin A levels were measured and the effects of two regimens of vitamin A supplementation on the incidence of BPD were compared. There were 24 infants in each of two groups with a mean birth weight less than 1000 g, a mean gestational age of 27 weeks, and similar sexual and racial mixes. In group I, vitamin A levels were measured after 1 week of parenteral nutrition and, if low, the infant was given a 2000-IU supplement intramuscularly three times a week, beginning on the 10th to 14th day of life. In group II, the same supplementation was begun on the second to fourth day of life. In both groups, when enteral feedings reached 60 kcal/kg per day, 2500 IU/d vitamin A was given orally. The incidence of oxygen support at 28 days was similar in both groups, although in group II there was a trend toward less vigorous ventilatory support. chi 2 analysis showed that a significantly smaller proportion of infants in group II had BPD at 36 weeks' gestational age than in group I. Length of neonatal intensive care unit stay was significantly reduced from 81 to 60 days. We conclude that vitamin A supplementation should be administered early to small, premature infants who are at risk for BPD and that monitoring of plasma levels is essential.