Abstract Context Blood cutoff values for vitamin A deficiency in children aged 3–10 years have not been addressed in the literature. Objective To identify blood retinol concentrations for determining severe vitamin A deficiency in children aged 3–10 years. Data Sources The MEDLINE, Web of Science, Embase, and Scopus databases were searched. Data Extraction Two reviewers independently extracted article data and assessed quality. Data Analysis The hierarchical summary receiver operating characteristic models were applied for the diagnostic accuracy meta-analysis. This review is registered at PROSPERO (identifier: CRD42020149367). Results A total of 15 articles met the eligibility criteria, and 9 were included in the diagnostic accuracy meta-analysis. The summary estimates (95%CI) were: Sensitivity, 0.39 (0.20–0.62); specificity, 0.79 (0.65–0.88); positive likelihood ratio, 1.85 (1.33–2.57); and negative likelihood ratio, 0.77 (0.60–0.99). The area under the curve of the overall analysis was 0.68 (95%CI 0.63–0.72). Conclusions Blood retinol concentrations have low diagnostic accuracy for severe vitamin A deficiency in children aged 3–10 years. Therefore, there is unclear evidence about the preferable cutoff point for determining severe vitamin A deficiency in children in this age group.