Abstract Background Raman spectroscopy (RS) has shown promise in the diagnosis of tuberculosis (TB). However, the evidence for its diagnostic accuracy in TB is still unclear. Results This meta-analysis incorporated a total of eight articles. RS exhibited a sensitivity of 0.91 (95% CI, 0.82-0.96) and specificity of 0.94 (95% CI, 0.85-0.97) for TB diagnosis. Importantly, the DOR was 146.18 (95% CI, 42.28-505.42), indicating a favorable discriminative ability. Additionally, the AUC of summary receiver operating characteristic curves was 0.97 (95% CI, 0.95-0.98). Subgroup analysis further indicated that both surface-enhanced Raman spectroscopy (SERS) and the 700-900 nm wavelength group demonstrated even higher sensitivity and specificity, surpassing 0.95. Significant heterogeneity was observed among the included studies; however, the publication bias was considered acceptable based on Deeks’ funnel plot analysis. Conclusions This meta-analysis suggests that RS may be an effective and accurate tool for differentiating TB. Notably, the application of SERS and the utilization of the 700-900 nm wavelength range further enhance the sensitivity and specificity. Trial Registration This meta-analysis has been registered at PROSPERO with ID: CRD42023415761.