Background: The effect of the hospital environment on health and specifically neurodevelopment in preterm infants remains under debate. We assessed outcomes of preterm infants hospitalised in single family rooms compared with common open bay units. Methods: For this systematic review and meta-analysis, we searched MEDLINE, Embase, PsycInfo, CENTRAL (the Cochrane Central Register of Controlled Trials), Web of Science, and ClinicalTrials.gov from inception to Aug 13, 2018, using controlled terms (ie, MeSH terms) and text words related to prematurity and neonatal intensive care unit design. We included randomised and non-randomised studies investigating clinical outcomes of preterm infants. We assessed methodological quality using the Cochrane Collaboration's Risk of Bias Tool for randomised controlled trials and the Cochrane Risk of Bias Tool for Non-randomised Studies of Interventions. We calculated summary estimates for meta-analysis using random effects models. The primary outcome was age appropriate long-term neurodevelopment. Secondary outcomes were length of hospital stay, sepsis, breastfeeding, growth, bronchopulmonary dysplasia, intraventricular haemorrhage, retinopathy of prematurity, and mortality. This systematic review is registered with PROSPERO, number CRD42016050643. Findings: We identified 487 records. 13 study populations (n=4793) were included. No difference in cognitive neurodevelopment was found on the Bayley Scales of Infant and Toddler Development-III at 18–24 months of corrected age (680 infants analysed; mean difference 1·04 [95% CI −3·45 to 5·52], p=0·65; I 2 =42%). The incidence of sepsis was lower (4165 infants analysed; 108 035 days in hospital [hospitalisation days]; risk ratio 0·63 [95% CI 0·50 to 0·78], p<0·0001; I 2 =0%) and exclusive breastfeeding at discharge was higher (484 infants analysed; 1·31 [1·07 to 1·61], p=0·01; I 2 =0%) in single family rooms than in open bay units. We found no differences in length of hospital stay, growth, bronchopulmonary dysplasia, intraventricular haemorrhage, retinopathy of prematurity, and mortality. Interpretation: Single family rooms should be considered to hospitalise preterm infants because incidence of sepsis is reduced and exclusive breastfeeding is higher. No difference in long-term neurodevelopment was detected. Funding: None.