Background <p>Red reflex testing for newborn screening detects only a fifth of disorders requiring intervention. In this systematic review and meta-analysis, we assessed the prevalence of sight-threatening disorders detected by screening newborn babies with retinal examination.</p> Methods <p>We searched PubMed, Embase, Web of Science and Scopus databases for studies assessing newborn retinal examination and reporting treatable sight-threatening abnormalities (cataract, glaucoma and retinoblastoma) and treatable potentially sight-threatening abnormalities (retinitis, vitreous hemorrhage, ROP-like retinopathy and familial exudative vitreoretinopathy). We excluded studies on preterm neonates reporting only ROP and on term neonates reporting only retinal hemorrhages. We adhered to MOOSE guidelines and used the JBI tool for quality assessment and R software for analysis (double arcsine transformation model).</p> Results <p>Thirteen observational studies (250,513 neonates) published between 2015 and 2024 were included in the meta-analysis. The studies were from China, India, Brazil, Malaysia, Indonesia, Turkey and New Zealand. Eleven studies used wide-field retinal imaging by RetCam and its variants, and one study each used indirect ophthalmoscopy and smartphone-based fundus imaging devices. On quality assessment, five studies were at increased risk of bias due to small sample size and/or not including consecutive births. The prevalence of treatable sight-threatening abnormalities was 6.1 per 10,000 neonates (13 studies, 250,513 neonates, 95% CI: 1, 14.1), and the prevalence of treatable potentially sight-threatening abnormalities was 42.2 per 10,000 neonates (12 studies, 250,167 neonates, 95% CI: 19.8, 71.9). No adverse effects were reported.</p> Conclusions <p>Eye screening of neonates with retinal examination can detect several sight-threatening abnormalities amenable to intervention.</p> Trial registration <p>PROSPERO database with registration number CRD42023448129.</p>