Objective The aim of the study was to summarize the diagnostic value of miR-21 as a biomarker in oral squamous cell carcinoma (OSCC) using a review of the literature and data from the cancer genome atlas (TCGA) database. Methods Data from TCGA database was sorted and analyzed by bioinformatics to determine the expression level of miR-21 in OSCC. Further, we searched for relevant articles in Embase, PubMed/Medline, Scopus, and Web of Science published before March 2021, extracted the data, and conducted quality assessment. The bivariate meta-analysis model with Stata 16.0 was used to analyze the diagnostic value of miR-21 for OSCC. Results A total of 304 related articles were identified, and seven were selected for meta-analysis. The diagnostic results after analysis were as follows: sensitivity 0.76 [95％ confidence interval (CI), 0.57–0.88]; specificity 0.77 (95％ CI, 0.58–0.89); positive likelihood ratio 3.34 (95％ CI, 1.58–7.08); negative likelihood ratio 0.31 (95％ CI, 0.15–0.63); diagnostic odds ratio 10.75 (95％ CI, 2.85–40.51); and area under the curve 0.83 (95％ CI, 0.80–0.86). The Deeks\' funnel chart showed that there was no potential bias (P = 0.54). Prediction analysis of the potential target genes of miR-21 was performed via the biological website, and DAVID was used to cross target genes for gene ontology (GO) annotation function analysis. Conclusion The results showed that miR-21-3p and miR-21-5p were significantly more highly expressed in OSCC tissues than in normal tissues (P < 0.05), and the results of the meta-analysis indicated that they could be used as potential biomarkers in the diagnosis of OSCC. In addition, 58 potential target genes of miR-21 were significantly enriched in 28 GO annotation functional pathways, which provided a biological basis for further clinical diagnostic value research.