PurposeTo synthesize evidence of mode of birth in extremely preterm vertex infants.MethodsWe searched MEDLINE, EMBASE, CINAHL, Cochrane CENTRAL, Web of Science, and ClinicalTrials.gov (January 1995–July 2019). We included studies comparing outcomes by vaginal birth or caesarean section in vertex infants born < 28 weeks or with birthweight < 1500 g. Two reviewers independently performed all steps. Primary outcomes were death and/or severe brain injury. We synthesised the data using random-effects meta-analyses (PROSPERO CRD42017074145).ResultsWe included 14 studies with 129,475 infants. In vertex singletons < 28 weeks, caesarean section was associated with reduced adjusted odds of death (aOR 0.62, 95% confidence interval [CI] 0.39–0.99, 3 studies, 10,331 infants). For severe brain injury or a composite of death or severe brain injury, adjusted data were lacking. In infants with very low birth weight overall (< 1500 g) we found no significant benefit for our primary outcomes (e.g., death, aOR 0.77, 0.55–1.07, 2 studies, 105,439 infants), although there were some benefits associated with caesarean section in smaller weight subgroups (e.g., death aOR 500–700 g: 0.53, 0.49–0.57 [1 study, 5989 infants] and 1000–1250 g: 0.78, 0.65–0.93 [1 study, 14,906 infants]), but not larger weights (1250–1500 g: 1.38, 1.15–1.65 [1 study, 17,715 infants]).ConclusionCaesarean section was associated with a significant decrease in the adjusted odds of death in extremely preterm vertex infants < 28 weeks. Smaller birth weight subgroups supported these results. The absence of randomized trials warrants judicious interpretation of these results, which are the currently available highest level of evidence. This study will inform further research.