Despite the improvements in the global fight against tuberculosis (TB), critical points still remain and fuel the epidemic. Even today, only 30% of the estimate number of people suffering from TB worldwide are correctly diagnosed, and lower proportions of cases are diagnosed in high-TB-burden, low-resource settings. Current TB diagnostics are still suboptimal in their performance for childhood TB, smear-negative TB, extrapulmonary TB, HIV-TB and drug-resistant TB. Furthermore, there is no gold standard test for the identification of latent TB infection status. Improving diagnosis is therefore a strategic goal in TB research, and the pipeline of diagnostic tools is rapidly growing: new ways of performing "old" tests (e.g. sputum smear microscopy) and completely innovative tools (e.g. new technologies for molecular diagnosis) are under investigation or have already been endorsed by WHO. Some of the structural limits of current TB diagnostics are likely to be overcome by such new tools, but research is still needed. Finally, the roll-out of new technologies and the development of newer ones will necessarily have to take into account the diagnostic needs of each context they are directed to (point-of-need testing approach), together with the logistic, economic and technical constraints present in the majority of high-TB-burden settings.