Background In women with threatened preterm birth, delay of delivery by 48 h allows antenatal corticosteroids to improve neonatal outcomes. For this reason, tocolytics are often administered for 48 h; however, there is no consensus about which drug results in the best maternal and neonatal outcomes. In the APOSTEL III trial we aimed to compare the effectiveness and safety of the calcium-channel blocker nifedipine and the oxytocin inhibitor atosiban in women with threatened preterm birth. Methods We did this multicentre, randomised controlled trial in ten tertiary and nine teaching hospitals in the Netherlands and Belgium. Women with threatened preterm birth (gestational age 25-34 weeks) were randomly assigned (1:1) to either oral nifedipine or intravenous atosiban for 48 h. An independent data manager used a web-based computerised programme to randomly assign women in permuted block sizes of four, with groups stratified by centre. Clinicians, outcome assessors, and women were not masked to treatment group. The primary outcome was a composite of adverse perinatal outcomes, which included perinatal mortality, bronchopulmonary dysplasia, sepsis, intraventricular haemorrhage, periventricular leukomalacia, and necrotising enterocolitis. Analysis was done in all women and babies with follow-up data. The study is registered at the Dutch Clinical Trial Registry, number NTR2947. Findings Between July 6, 2011, and July 7, 2014, we randomly assigned 254 women to nifedipine and 256 to atosiban. Primary outcome data were available for 248 women and 297 babies in the nifedipine group and 255 women and 294 babies in the atosiban group. The primary outcome occurred in 42 babies (14%) in the nifedipine group and in 45 (15%) in the atosiban group (relative risk [RR] 0·91, 95% CI 0·61-1·37). 16 (5%) babies died in the nifedipine group and seven (2%) died in the atosiban group (RR 2·20, 95% CI 0·91-5·33); all deaths were deemed unlikely to be related to the study drug. Maternal adverse events did not differ between groups. Interpretation In women with threatened preterm birth, 48 h of tocolysis with nifedipine or atosiban results in similar perinatal outcomes. Future clinical research should focus on large placebo-controlled trials, powered for perinatal outcomes. Funding ZonMw (the Netherlands Organisation for Health Research and Development).