Objective To systematically evaluate the value of CXC chemokine ligand 9 (CXCL9) in the diagnosis of tuberculosis. Methods Literature on the diagnosis of tuberculosis by plasma CXCL9 were collected by searching PubMed, EBSCO, China National Knowledge Infrastructure and Wanfang databases form establishment to September 2022, and then literature screening, data extraction and quality assessment were performed independently by two researchers. Meta-analysis was performed using Review Manager 5.3 and Stata 15 softwares. Results According to the inclusion and exclusion criteria, 10 articles were selected, including 1586 participants from 5 countries and regions. Meta-analysis results showed that the positive rate of CXCL9 was higher in active tuberculosis patients than that in healthy people and latent tuberculosis patients [active tuberculosis patients vs. healthy people: odds ratio (OR)=21.69, 95% confidence interval (CI) (6.52, 72.16), P<0.00001; active tuberculosis patients vs. latent tuberculosis patients: OR=10.12, 95%CI (3.83, 26.76), P<0.00001]. The sensitivity and specificity of plasma CXCL9 in distinguishing active tuberculosis patients from healthy people were 0.84 [95%CI (0.77, 0.90)] and 0.82 [95%CI (0.63, 0.92)], respectively; the sensitivity and specificity of plasma CXCL9 in distinguishing active tuberculosis patients from latent tuberculosis patients were 0.77 [95%CI (0.56, 0.90)] and 0.72 [95%CI (0.40, 0.91)], respectively. Subgroup analysis showed that the infection status of human immunodeficiency virus had some impact on heterogeneity, while other factors had limited impact on heterogeneity. Egger test showed that there was no publication bias (active tuberculosis patients vs. healthy people: P=0.976; active tuberculosis patients vs. latent tuberculosis patients: P=0.606). Conclusion CXCL9 has a high diagnostic value for tuberculosis patients and may be used as a new biomarker to diagnose tuberculosis.