Tuberculosis diagnosis often involved a variety of methods or examinations that makes it possible to conduct tuberculosis diagnostic accuracy studies based on routine work. However, most of the current diagnostic accuracy of tuberculosis research still have quality problems, such as patients’ selection, gold standard verification, blind evaluation, and so on. All these requires researchers to rigorous design, standardized report. So, this paper systematically introduces the design of diagnostic accuracy, indicators analysis and some problems need to pay attention, and help researchers to carry out high-quality original research.