Nasal masks and nasal prongs are used as interfaces for providing continuous positive airway pressure (CPAP) for preterm infants with or at risk of respiratory distress, either as primary support after birth or as ongoing support after endotracheal extubation from mechanical ventilation. It is unclear which type of interface is associated with lower rates of CPAP treatment failure, nasal trauma, or mortality and other morbidity. To assess the benefits and harms of nasal masks versus nasal prongs for reducing CPAP treatment failure, nasal trauma, or mortality and other morbidity in newborn preterm infants with or at risk of respiratory distress. We used standard, extensive Cochrane search methods. The latest search date was October 2021. We included randomised controlled trials comparing masks versus prongs as interfaces for delivery of nasal CPAP in newborn preterm infants (less than 37 weeks' gestation) with or at risk of respiratory distress. We used standard Cochrane methods. Our primary outcomes were 1. treatment failure, 2. all-cause mortality, and 3. neurodevelopmental impairment. Our secondary outcomes were 4. pneumothorax, 5. moderate-severe nasal trauma, 6. bronchopulmonary dysplasia, 7. duration of CPAP use, 8. duration of oxygen supplementation, 9. duration of hospitalisation, 10. patent ductus arteriosus receiving medical or surgical treatment, 11. necrotising enterocolitis, 12. severe intraventricular haemorrhage, and 13. severe retinopathy of prematurity. We used the GRADE approach to assess the certainty of the evidence. We included 12 trials with 1604 infants. All trials were small (median number of participants 118). The trials occurred after 2001 in care facilities internationally, predominantly in India (eight trials). Most participants were preterm infants of 26 to 34 weeks' gestation who received nasal CPAP as the primary form of respiratory support after birth. The studied interfaces included commonly used commercially available masks and prongs. Lack of measures to blind caregivers or investigators was a potential source of performance and detection bias in all the trials. Meta-analyses suggested that use of masks compared with prongs may reduce CPAP treatment failure (risk ratio (RR) 0.72, 95% confidence interval (CI) 0.58 to 0.90; 8 trials, 919 infants; low certainty). The type of interface may not affect mortality prior to hospital discharge (RR 0.83, 95% CI 0.56 to 1.22; 7 trials, 814 infants; low certainty). There are no data on neurodevelopmental impairment. Meta-analyses suggest that the choice of interface may result in little or no difference in the risk of pneumothorax (RR 0.93, 95% CI 0.45 to 1.93; 5 trials, 625 infants; low certainty). Use of masks rather than prongs may reduce the risk of moderate-severe nasal injury (RR 0.55, 95% CI 0.44 to 0.71; 10 trials, 1058 infants; low certainty). The evidence is very uncertain about the effect on bronchopulmonary dysplasia (RR 0.69, 95% CI 0.46 to 1.03; 7 trials, 843 infants; very low certainty). The available trial data provide low-certainty evidence that use of masks compared with prongs as the nasal CPAP interface may reduce treatment failure and nasal injury, and may have little or no effect on mortality or the risk of pneumothorax in newborn preterm infants with or at risk of respiratory distress. The effect on bronchopulmonary dysplasia is very uncertain. Large, high-quality trials would be needed to provide evidence of sufficient validity and applicability to inform policy and practice.