Background: A multicenter randomized trial comparing the effects of dex therapy initiated at 14 or 28 days of life in VLBW (501-1500g) infants at risk for chronic lung disease (CLD) was conducted by the NICHD Neonatal Research Network. Group 1 received dex on study days 1-14 and placebo on study days 14-28. Group 2 received placebo on study days 1-14 and steroids, if needed, on study days 14-28. Group 2 infants with minimal or no ventilator dependence on study day 14 continued to receive placebo to study day 28.Purpose: To assess the relation of dex and other factors to definite or possible/clinical infection. Methods: Data on suspicion of infection, cultures performed, and antibiotic therapy were collected daily for 42 days after randomization. Infectious complications were evaluated for study days 1-14 and 1-42. Findings: 371 patients were enrolled. Group 1 (n=182) and Group 2 (n=189) had similar baseline risk factors for infection. During the first 14 days, Group 1 infants were significantly more likely than Group 2 infants to have a positive blood culture (36% vs 24%, p=0.012) and a definite infection (22% vs 14%, p=0.038). Over the 6-week study period, 42% of all infants had at least one positive blood culture (Group 1, 46% vs. Group 2, 38%, p=0.12) and 25% of infants were judged to have one or more definite infections (Group 1, 29% vs. Group 2, 21%, p=0.08). Among patients who were cultured, infants in Group 1 were significantly more likely to have a positive blood culture than were those in Group 2 (53% vs. 41%, p=0.02). Factors present at the time of randomization (including BW, GA, race, sex, prenatal steroids, C/S, sepsis hx, severity of RDS, medications given and treatment assignment) were evaluated for their association with subsequent infection. Using logistic regression models, randomization to Group 1 and antacid/H2 blocker therapy were both associated with an increased risk of definite infection for days 1-14 and antacid/H2 blocker treatment increased risk for the entire study period. Increasing birthweight and C/S delivery were both associated with a decreased risk of definite infection.Conclusion: Contrary to the findings of smaller trials, infection was more common in patients randomized to dex than placebo during the first 2 study weeks. In addition, the risk of infection over the entire 42-day study period was higher when dex was started at 2 weeks of age than when dex was started at 4 weeks of age, if needed. The decision to treat VLBW infants with or at risk for CLD with dex must be based on a balance of potential benefits and risks.