ObjectiveThe diagnostic efficacy of long non-coding RNA (lncRNA) for tuberculosis was evaluated by systematic review. MethodsData from PubMed, Web of Science, Cochrane Library, Embase, CMJFD, CNKI and WanFang Data were searched. Literatures on the diagnostic value of lncRNA in tuberculosis from the database establishment to August 20, 2024 were selected, and the quality of literatures was assessed using QUADAS-2 tool. Meta-Disc 1.0 software tested the threshold heterogeneity of the included studies. Stata 18.0 software calculated sensitivity, specificity, positive likelihood ratio, negative likelihood ratio, diagnostic odds ratio and other effect sizes, and performed subgroup analysis and meta regression to explore the source of heterogeneity. Deeks funnel plot evaluates publication bias. Results A total of 28 case-control studies were included in 14 literatures. The meta-results showed that the combined sensitivity was 0.88 (95%CI 0.81 to 0.93), the specificity was 0.90 (95%CI 0.84 to 0.94), and the PLR was 9.05 (95%CI 5.16 to 15.87). The NLR and DOR were 0.13 (95%CI 0.08 to 0.22) and 67.96 (95%CI 27.27 to 169.39), and the AUC were 0.95 (95%CI 0.93 to 0.97). Subgroup analysis showed that lncRNA was more effective in the diagnosis of tuberculosis when PMBC samples, lncRNA expression was down-regulation, the study sample size was ≤100, there was cut-off value, GAPDH was used as the internal reference, and RNA extraction kit was used. meta regression indicated that lncRNA expression level and sample size were the main sources of heterogeneity. Conclusion LncRNA has high accuracy in the diagnosis of tuberculosis, and is expected to become a new biomarker to assist the diagnosis of tuberculosis.