Efforts to control tuberculosis are currently hampered by the lack of effective tools for the detection of infected individuals. Although new diagnostic tests are being developed and marketed, there is a paucity of data on the value of new technology for the diagnosis of tuberculosis in endemic regions of the world. Trials of new tools need to be undertaken in appropriate settings and to a high standard. The value of previous studies has frequently been compromised by inadequate study design and poor execution. We offer guidelines to assist in planning trials for new tests for the diagnosis of tuberculosis.