This systematic review assesses the accuracy of molecular diagnostic methods for the detection of pulmonary tuberculosis in studies performed in China, published in Chinese and English. We searched for studies that assessed the accuracy of molecular diagnostics for pulmonary TB in China in the China National Knowledge Infrastructure, the Wanfang Database, SinoMed, VIP Information, Pubmed, Embase, and the Cochrane Library. For each index test, a summary estimation for sensitivity and specificity was calculated using the bivariate random-effects model. A total of 59 studies were included in our analysis. Loop-mediated isothermal amplifcation (LAMP) assay (six studies; pooled sensitivity 90%, 95% CI 78-95%; specificity 93%, 85-97%), line probe assay (LPA) (one study; 87%, 84-90%; 94%, 92-95%) and polymerase chain reaction (PCR) (FQ-PCR and RT-PCR) (four studies; 90%, 55-99%; 93%, 71-99%) showed good diagnostic performance in the meta-analysis. The highest pooled sensitivity was from Xpert MTB/RIF (20 studies; pooled sensitivity 91%, 95% CI 87-94%). The highest pooled specificity was from cross-priming amplification (CPA) (six studies; pooled specificity 97%, 95-99%). The lowest pooled sensitivity and specificity were from simultaneous amplification and testing (SAT)-TB (three studies; 79%, 66-88%; 72%, 48-88%). In subgroup analysis, molecular diagnostics demonstrated higher sensitivity for pulmonary TB detection in smear-positive specimens. Xpert MTB/RIF, LAMP, LPA, CPA and PCR demonstrated high accuracy overall for pulmonary tuberculosis detection, while SAT-TB had poor performance.