Bronchopulmonary dysplasia is a chronic, sometimes fatal lung disease, which primarily affects premature infants and often leads to a dependence on mechanical ventilation lasting many months. To identify prognostic factors of mortality at 1 and 2 months of age, the authors reviewed the medical records of the 144 neonates admitted to two neonatal intensive care units in Seattle from January 1, 1986, through December 31, 1988, who required mechanical ventilation throughout the first month of life. Likely predictors of mortality were tested by logistic regression analysis. The calculated mean airway pressure at 30 days of age (MAP30) and the diagnosis of bacterial sepsis at any time during the first month of life (Bact0-30) were statistically significant predictors of mortality (P < .001 and P = .018, respectively) and had the lowest deviance in the regression model. The probability of mortality was estimated by 1/(1 + e-x, where x = -6.510 + 0.4588 (MAP30) + 1.475 (Bact0-30), and where MAP30 is expressed as centimeters of water pressure (1 cm H2O = 0.0978 kPa) and the presence or absence of bacteremia is 1 and 0, respectively. The records of the 57 infants who still required mechanical ventilation at 60 days of age were reanalyzed with clinical data available during the first 2 months of life. Mean airway pressure (MAP60) and the fraction of inspired oxygen (F60) at 60 days of age combined to form the best predictors of mortality, where x = -7.668 + 0.2940 (MAP60) + 5.935 (F60). The occurrence of bacterial sepsis during the first 2 months of life, the degree of hypochloremia, and the duration of chronic sedative use were also significant predictors of survival, even controlling of MAP60 and F60. These regression equations allow more accurate estimation of the likelihood of survival for chronically ventilated infants and may facilitate decisions regarding withdrawal or continuation of life support.