OBJECTIVE. The goal was to develop multivariate logistic regression models for the outcome of bronchopulmonary dysplasia and/or death at postmenstrual age of 36 weeks by using clinical and cytokine data from the first 28 days. METHODS. For 1067 extremely low birth weight infants in the Neonatal Research Network of the National Institute of Child Health and Human Development, levels of 25 cytokines were measured in blood collected within 4 hours after birth and on days 3, 7, 14, and 21. Stepwise regression analyses using peak levels of the 25 cytokines and 15 clinical variables identified variables associated with bronchopulmonary dysplasia/death. Multivariate logistic regression analysis was performed for bronchopulmonary dysplasia/death by using variables selected through stepwise regression. Similar analyses were performed by using average cytokine values from days 0 to 21, days 0 to 3, and days 14 to 21. RESULTS. Of 1062 infants with available data, 606 infants developed bronchopulmonary dysplasia or died. On the basis of results from all models combined, bronchopulmonary dysplasia/death was associated with higher concentrations of interleukin 1β, 6, 8, and 10 and interferon γ and lower concentrations of interleukin 17, regulated on activation, normal T cell expressed and secreted, and tumor necrosis factor β. Compared with models with only clinical variables, the addition of cytokine data improved predictive ability by a statistically significant but clinically modest magnitude. CONCLUSIONS. The overall cytokine pattern suggests that bronchopulmonary dysplasia/death may be associated with impairment in the transition from the innate immune response mediated by neutrophils to the adaptive immune response mediated by T lymphocytes. Copyright © 2009 by the American Academy of Pediatrics.