Background: Antenatal corticosteroid (ACS) therapy is well established for improving respiratory outcomes in early preterm infants; however, its role in late preterm neonates remains controversial because of uncertainties regarding real-world applicability, optimal timing, patient selection, and potential confounding. This study aimed to evaluate the adjusted association between ACS exposure and neonatal respiratory morbidity among late preterm infants. Methods: This retrospective single-center cohort study screened 1,786 late preterm births between January 2014 and December 2019. After application of strict eligibility criteria, 265 neonates (14.8%) born at 34⁰/₇ to 36⁶/₇ weeks of gestation were included in the final analysis. Neonates born to mothers who received a complete course of ACS and delivered between 24 h and 7 days after completion of therapy constituted the ACS-exposed group ( n = 112), while those without ACS exposure during the late preterm period constituted the non-exposed group ( n = 153). Respiratory outcomes were classified according to the level and duration of respiratory support. The primary outcome was composite severe respiratory morbidity. Multivariable logistic regression was used to estimate adjusted associations. Results: Overall respiratory morbidity was observed less frequently among ACS-exposed neonates. In adjusted analysis including gestational age as completed-week categories, neonates not exposed to ACS had higher odds of composite severe respiratory morbidity compared with ACS-exposed neonates (adjusted OR 2.18, 95% CI 1.11–4.27, p = 0.023). Conclusions: In this retrospective real-world cohort, complete ACS exposure within a defined 24-hour to 7-day ACS-to-delivery interval was associated with lower odds of severe respiratory morbidity among late preterm neonates. Because the analysis included 265 of 1,786 screened late preterm births and focused on neonates with complete ACS exposure within the predefined ACS-to-delivery interval, potential selection bias and limited generalizability should be considered. Given the observational design, gestational age imbalance, maternal comorbidity imbalance, and potential residual confounding, these findings should be interpreted as adjusted associations rather than evidence of a causal effect.