Objective We evaluated the best time to initiate delivery in late preterm pre-eclampsia in order to optimise long-term infant and maternal outcomes. Design Parallel-group, non-masked, randomised controlled trial. Setting 46 UK maternity units. Population Women with pre-eclampsia between 34+0 and 36+6 weeks’ gestation, without severe disease, were randomised to planned delivery or expectant management. Primary long-term outcome Infant neurodevelopmental outcome at 2 years of age, using the PARCA-R (Parent Report of Children’s Abilities-Revised) composite score. Results Between 29 September 2014 and 10 December 2018, 901 women were enrolled in the trial, with 450 allocated to planned delivery and 451 to expectant management. At 2-year follow-up, the intention-to-treat analysis population included 276 women (290 infants) allocated to planned delivery and 251 women (256 infants) to expectant management. The mean composite standardised PARCA-R scores were 89.5 [standard deviation (SD) 18.2] in the planned delivery group and 91.9 (SD 18.4) in the expectant management group, with an adjusted mean difference of −2.4 (95% CI −5.4 to 0.5) points. Conclusion In infants of women with late preterm pre-eclampsia, average neurodevelopmental assessment at 2 years lies within the normal range, regardless of whether planned delivery or expectant management is pursued. Because of lower than anticipated follow-up, there was limited power to demonstrate these scores were not different, but the small between-group difference in PARCA-R scores is unlikely to be clinically important.