ABSTRACT This study aimed to compare the performance of stool-based Xpert MTB/RIF(Xpert) assay (Cepheid, USA) against that of smear, culture, and Xpert using respiratory tract specimens (RTS) and stool samples in the diagnosis of adult pulmonary tuberculosis (PTB). This prospective study of patients with presumptive PTB was conducted at the Beijing Chest Hospital from January 2021 to December 2021. The smear test by fluorescence microscope, MGIT 960 liquid culture, and Xpert MTB/RIF assay was performed simultaneously on the RTS and stool samples. Patients were grouped based on the RTS examination results and clinical diagnosis. A total of 434 eligible patients were enrolled. Among these, 351 patients were diagnosed as PTB, while 83 patients were diagnosed as non-TB patients. Using RTS as standard sample, the sensitivity of smear on stool, culture on stool, Xpert on stool, and Xpert on RTS was 13.39% (47/351), 22.22% (78/351), 45.30% (159/351), and 66.10% (232/351), respectively. Furthermore, the specificity of Xpert on RTS and stool was 100% (83/83) for both specimens. Among the 48 patients with PTB confirmed by bronchoalveolar lavage fluid (BALF) examination, 16 had stool specimens with positive Xpert results. Similarly, among 200 patients diagnosed through sputum examination, 133 had positive Xpert results on stool specimens. Thus, Xpert on stool specimens may be a promising and practical strategy to improve PTB diagnosis, particularly among patients unable to expectorate sputum.IMPORTANCEThis study aimed to assess the value of the Xpert MTB/RIF assay (Xpert) for diagnosing PTB using stool samples from adults in low HIV prevalence settings. Although the diagnostic yield of Xpert on stool samples is inferior to that of Xpert on RTS, stool-based Xpert may be useful in diagnosing patients with presumptive PTB who cannot expectorate sputum and do not opt for BALF collection.