The incidence of culture-proven neonatal sepsis and necrotizing enterocolitis (NEC) in preterm infants maintained at pharmacologic (mean 5.1 mg/dL +/- 1.45 SD) serum vitamin E levels for long periods was prospectively studied as part of a double-masked clinical trial of the effect of prophylactic vitamin E v placebo treatment on the development and course of retinopathy of prematurity (ROP). Within a few days of birth, 914 preterm infants were enrolled in the study; 545 (275 placebo-treated infants, 270 vitamin E-treated infants had birth weight of 1,500 g or less. A significant difference in incidence of neonatal sepsis (17 placebo-treated infants, 37 vitamin E-treated infants) and NEC (18 placebo-treated infants, 32 vitamin E-treated infants) was observed among infants who had been treated for eight or more days and who had developed neither sepsis nor NEC before that time. The association of vitamin E treatment with increased incidence of disease was much higher with sepsis than with NEC. The most likely reason for these observations is a pharmacologic serum vitamin E-related decrease in oxygen-dependent intracellular killing ability which results in a decreased resistance to infection in preterm infants. The data suggest that, if this occurs, it is clinically significant only in the more immature infants. In view of the known variability of absorption of oral vitamin E and the association between high serum vitamin E levels and increased incidence of sepsis and late-onset NEC reported here, it can be concluded that serum vitamin E levels must be monitored when supplemental vitamin E is administered to premature infants, especially those with birth weight 1,500 g or less. The risk-benefit ratio of long-term treatment using vitamin E at high serum levels should be clearly assessed.