ObjectiveTo quantify trends in caffeine use in infants born at <32 weeks’ gestational age (GA), and to investigate the effects of early vs late caffeine on neonatal outcomes.Study designRetrospective propensity score matched cohort study using routinely recorded data from the National Neonatal Research Database of infants born at <32 weeks’ GA admitted to neonatal units in England and Wales (2012–2020).Results89% (58 913/66 081) of infants received caffeine. In 70%, caffeine was started early (on the day of birth or the day after), increasing from 55% in 2012 to 83% in 2020. Caffeine was given for a median (IQR) of 28 (17–43) days starting on day 2 (1–3) and continued up to 34 (33–34) weeks postmenstrual age.In the propensity score matched cohort of 13 045 pairs of infants, the odds of preterm brain injury (early caffeine, 2306/13 045 (17.7%) vs late caffeine, 2528/13 045 (19.4%), OR=0.89 (95% CI 0.84 to 0.95)) and bronchopulmonary dysplasia (BPD) (early caffeine, 4020/13 045 (32.8%) vs late caffeine, 4694/13 045 (37.7%), OR=0.81 (95% CI 0.76 to 0.85)) were lower in the group that received early caffeine compared with those who received it later.ConclusionsEarly use of caffeine has increased in England and Wales. This is associated with reduced risks of BPD and preterm brain injury. Randomised trials are needed to find the optimal timing of caffeine use and the groups of infants who will benefit most from early administration of caffeine.