Purpose: Leukoencephalopathy with vanishing white matter (VWM) is an autosomal recessive leukoencephalopathy caused by mutations in any of the five genes encoding the subunits of eukaryotic translation initiation factor 2B (eIF2B). The severity of the disease varies considerably, and its genotypic-phenotypic correlation is still unclear. Age of onset is the only independent clinical predictor for VWM severity. In this study, the correlation between genotype and age at onset of patients was investigated. Methods: Data were collected from patients with VWM in the available literature reports and from those diagnosed in Peking University First Hospital. The age of onset was divided into early-onset (≤4 years) and late-onset type (>4 years) for the analysis of the correlation between genotype and age of onset in patients with VWM. Results: A total of 341 patients were included, 281 were reported in 87 available articles and 60 were diagnosed in our center. A total of 180 different mutations were found, among which 86.1% were missense. The gene (EIF2B1-5) in which the mutation located, and the number of null alleles were not associated with age of onset in these patients. Certain mutations such as eIF2Bε[Arg195His] and eIF2Bε[Arg269Gln] that were predicted to have a serious influence on eIF2B structure were related to earlier age of onset. EIF2Bγ[Ala87Val] which was predicted to have a minimal influence on eIF2B structure, was related to later age of onset. Whereas eIF2Bß[Glu213Gly], eIF2Bß[Gly200Val] and eIF2Bε[Thr91Ala], also predicted having a small effect on the structure of eIF2B, did not show correlation with the age of onset. The onset age of patients with one or biallelic missense mutations located in the catalytic domain or other homologous domains in catalytic subunits (eIF2Bγ, ε) was earlier than that of patients with biallelic mutations located in the NT domain. Conclusion: The onset age of patients with different genotypes varied greatly. The degree of influence in protein structure of some missense mutations was correlated with phenotypic severity, but the results were not completely consistent. The combined effect of biallelic mutations, the role of regulatory genes, environmental stress and other potential factors on phenotypes need to be further explored.