OBJECTIVE: Infant boys have worse outcomes than girls. In twins, the 'male disadvantage' has been reported to extend to female co-twins via a 'masculinising' effect. We studied the association between sex pairing and neonatal outcomes in extremely preterm twins. DESIGN: Retrospective cohort study SETTING: Eleven countries participating in the International Network for Evaluating Outcomes of Neonates. PATIENTS: Liveborn twins admitted at 23-29 weeks' gestation in 2007-2015. MAIN OUTCOME MEASURES: We examined in-hospital mortality, grades 3/4 intraventricular haemorrhage or cystic periventricular leukomalacia (IVH/PVL), bronchopulmonary dysplasia (BPD), retinopathy of prematurity requiring treatment and a composite outcome (mortality or any of the outcomes above). RESULTS: Among 20 924 twins, 38% were from male-male pairs, 32% were from female-female pairs and 30% were sex discordant. We had no information on chorionicity. Girls with a male co-twin had lower odds of mortality, IVH/PVL and the composite outcome than girl-girl pairs (reference group): adjusted OR (aOR) (95% CI) 0.79 (0.68 to 0.92), 0.83 (0.72 to 0.96) and 0.88 (0.79 to 0.98), respectively. Boys with a female co-twin also had lower odds of mortality: aOR 0.86 (0.74 to 0.99). Boys from male-male pairs had highest odds of BPD and composite outcome: aOR 1.38 (1.24 to 1.52) and 1.27 (1.16 to 1.39), respectively. CONCLUSIONS: Sex-related disparities in outcomes exist in extremely preterm twins, with girls having lower risks than boys and opposite-sex pairs having lower risks than same-sex pairs. Our results may help clinicians in assessing risk in this large segment of extremely preterm infants.