The aim of the present study was to collect published studies and compare the diagnostic accuracy of different markers for nasopharyngeal carcinoma (NPC). We systematically searched PubMed/MEDLINE, EMBASE, Cochrane Library, CNKI, and Wanfang for relevant studies until April 29, 2020. The revised Quality Assessment of Diagnostic Accuracy Studies (QUADAS-2) tool was used to evaluate the methodological quality of the studies. The sensitivity, specificity, positive likelihood ratio (PLR), negative likelihood ratio (NLR), diagnostic odds ratio (DOR), and area under the curve (AUC) values of the diagnostic markers were combined by a bivariate mixed effect model to compare their diagnostic accuracy. We explored heterogeneity through meta-regression. In total, 244 records from 101 articles were included, with 49,432 total study subjects (13,109 cases and 36,323 controls). EA-IgG, Zta-IgG, and Epstein-Barr virus (EBV) DNA load in non-invasive nasopharyngeal brushings (EBV-DNA brushings) have both high sensitivity and specificity, EBNA1-IgG and VCA-IgG have only high sensitivity, and EBNA1-IgA, VCA-IgA, Rta-IgG, Zta-IgA, HSP70, and serum sialic acid (SA) have only high specificity. The bivariate mixed effect model of EA-IgA had a significant threshold effect. Meta-regression analysis showed that ethnicity affected EBNA1-IgA, EBNA1-IgG, VCA-IgA, and EBV DNA load in plasma, test methods affected EBNA1-IgG, publication year affected VCA-IgA, and sample size affected Rta-IgG. There was significant publication bias for VCA-IgA and Rta-IgG (P < 0.05). EA-IgG, Zta-IgG, and EBV-DNA brushings are good diagnostic markers for NPC. The diagnostic accuracy was influenced by publication year, sample size, test methods, and ethnicity.