Bronchopulmonary dysplasia is the most common morbidity of prematurity, but the validity and utility of commonly used definitions have been questioned. To compare three commonly used definitions of bronchopulmonary dysplasia in a contemporary prospective, multicenter observational cohort of extremely preterm infants. At 36 weeks postmenstrual age, the following definitions of bronchopulmonary dysplasia were applied to surviving infants with and without imputation: need for supplemental oxygen (Shennan definition), National Institutes of Health Workshop definition, and "physiologic" definition after a room-air challenge. Of 765 survivors assessed at 36 weeks, bronchopulmonary dysplasia was diagnosed in 40.8, 58.6, and 32.0% of infants, respectively, with the Shennan, workshop and physiologic definitions. The number of unclassified infants was lowest with the workshop definition (2.1%) and highest with the physiologic definition (16.1%). After assigning infants discharged home in room air before 36 weeks as no bronchopulmonary dysplasia, the modified Shennan definition compared favorably to the workshop definition, with 2.9% unclassified infants. Newer management strategies with nasal cannula flows up to 4 L/min or more and 0.21 FiO2 at 36 weeks obscured classification of bronchopulmonary dysplasia status in 12.4% of infants. Existing definitions of bronchopulmonary dysplasia differ with respect to ease of data collection and number of unclassifiable cases. Contemporary changes in management of infants, such as use of high-flow nasal cannula, limit application of existing definitions and may result in misclassification. A contemporary definition of bronchopulmonary dysplasia that correlates with respiratory morbidity in childhood is needed. Clinical trial registered with www.clinicaltrials.gov (NCT01435187).