ABSTRACT: BACKGROUND:The current methods for detecting Mycobacterium tuberculosis (Mtb) are not clinically optimal. Standard culture methods (SCMs) are slow, costly, or unreliable, and loop-mediated isothermal amplification (LAMP) cannot differentiate live Mtb. METHODS:This study compared reverse transcription (RT)-LAMP, LAMP, and an SCM for detecting Mtb. A first experiment tested the sensitivity and specificity of primers for 9 species of Mycobacterium (H37Rv, M. intracellulare, M. marinum, M. kansasii, M. avium, M. flavescens, M. smegmatis, M. fortuitum, and M. chelonae); and 3 non-Mycobacterium species (Staphylococcus aureus, Pseudomonas aeruginosa, and Klebsiella pneumoniae). A second experiment tested sputum specimens for the presence of Mtb, from 100 patients with tuberculosis (clinical) and 22 from patients without tuberculosis (control), using Roche solid culture (SCM), LAMP, and RT-LAMP. In the clinical samples. RESULTS:The rates of positivity for Mtb of the SCM, LAMP, and RT-LAMP methods were 88%, 92%, and 100%, respectively. The difference in detection rate was significant between RT-LAMP and SCM, but RT-LAMP and LAMP were comparable. In the control group, the detection rates were nil for all three methods. CONCLUSION:The specificities of the methods were similar. The sensitivity of RT-LAMP was ~10-fold higher than that of LAMP for detecting Mtb. Unlike LAMP, RT-LAMP could identify viable bacteria, and was able to detect a single copy of Mtb. Among SCM, LAMP, and RT-LAMP, the latter is the most suitable for wide use in the lower-level hospitals and clinics of China for detecting Mtb in sputum samples.