Objectives: To assess the caffeine response in preterm and term newborns with persistent intermittent hypoxia (IH) as they approach discharge, with a particular focus on newborns with chronic lung disease of prematurity (CLD). To identify factors associated with a lack of response to caffeine. Design/settings: Retrospective cohort study across two neonatal intensive care units between 2015-2022. Main outcomes following caffeine administration: Normal oximetry; no need for respiratory support; resolution of hypercapnia. Results. A total of 132 infants received caffeine for persistent IH. Normal oximetry was achieved post-caffeine in 81% (46/57) of newborns with CLD, in 96% (46/48) of preterm with no CLD, in 96% (26/27) of term newborns. Caffeine reduced the % time with SpO 2 <90% from 6.8% (interquartile range, 3.8%-12.2%), to 0.8% (0.4%-1.6%, p <.0001). The desaturation index <80% dropped from 5.3 events/hour (0.9-14.6 events/hour) to 0.2 events/hour (0-0.78 events/hour, p <.0001) and the index ≥10% lasting >10 s went from 6.6 events/hour (3.3-10.7 events/hour) to 1.4 events/hour (0.7-2.4 events/hour, p <.0001). Of the 61 infants on respiratory support, 45 (74%) were weaned within a few days following caffeine. Caffeine normalized PCO 2 in 63% (41/65) of newborns with elevated PCO 2 pre-caffeine. Infants failing caffeine were more likely to have CLD compared to responders (79% vs. 39%, p<.005). Caffeine was successfully discontinued (first attempt, normal oximetry) in 101 infants (88%) at postmenstrual age of 46.1 weeks (45.3-48.3 weeks).Conclusion: Caffeine improved respiratory outcomes in the majority of preterm and term born infants with persistent IH, including those with CLD.