[Adapted from the doctoral dissertation submitted to Paris-Descartes University on November 26, 2019, by Ayoub Mitha, under the supervision of Véronique Pierrat and Pierre-Yves Ancel, "Breastfeeding at Discharge Among Preterm Infants and Unit Policies: Data from the EPIPAGE-2 Cohort."] Background - Despite the recognized benefits of breast milk, breastfeeding rates among preterm infants vary by country and neonatal unit. Maternal and neonatal characteristics associated with breastfeeding a preterm infant have been extensively studied but do not fully explain this variability. Objectives - To describe and analyze factors associated with breastfeeding at hospital discharge among preterm infants, with a particular focus on unit policies supporting breastfeeding. Methods - The Epipage-2 cohort (Epidemiological Study of Low Gestational Ages-2) is a national prospective population-based study that included births occurring in 2011 between 22 and 34 weeks of gestation (WG) in all neonatal units across 25 French regions. Determinants of breastfeeding were analyzed using multivariate logistic regression in two distinct populations: 3,108 infants born before 32 GA and 883 infants born between 32 and 34 GA. Results - In total, 47% of infants born before 32 weeks GA and 59% of infants born between 32 and 34 weeks GA were receiving breast milk at hospital discharge, with inter-unit variability ranging from 21% to 84% and from 27% to 87%, respectively. Unit policies supporting breastfeeding, particularly early skin-to-skin contact and parental involvement in feeding their child, were partially associated with this variability. High rates of breastfeeding initiation in the general population were associated with breastfeeding at discharge only among infants born between 32 and 34 weeks’ gestational age. Conclusion—Adopting the supportive policies of the highest-performing units could help reduce variability in breastfeeding rates at discharge among this at-risk population of preterm infants.