Anti-tuberculosis (TB) drugs possess diverse abilities to penetrate the different host tissues and cell types in which infecting Mycobacterium tuberculosis bacilli are located during active disease. This is important since there is increasing evidence that the respective "lesion-penetrating" properties of the front-line TB drugs appear to correlate well with their specific activity in standard combination therapy. In turn, these observations suggest that rational efforts to discover novel treatment-shortening drugs and drug combinations should incorporate knowledge about the comparative abilities of both existing and experimental anti-TB agents to access bacilli in defined physiological states at different sites of infection, as well as avoid elimination by efflux or inactivation by host or bacterial metabolism. However, while there is a fundamental requirement to understand the mode of action and pharmacological properties of any current or experimental anti-TB agent within the context of the obligate human host, this is complex and, until recently, has been severely limited by the available methodologies and models. Here, we discuss advances in analytical models and technologies which have enabled investigations of drug metabolism and pharmacokinetics (DMPK) for new TB drug development. In particular, we consider the potential to shift the focus of traditional pharmacokinetic-pharmacodynamic analyses away from plasma to a more specific "site of action" drug exposure as an essential criterion for drug development and the design of dosing strategies. Moreover, in summarising approaches to determine DMPK data for the "unit of infection" comprising host macrophage and intracellular bacillus, we evaluate the potential benefits of including these analyses at an early stage in the preclinical drug development algorithm. © 2018 IUBMB Life, 70(9):926-937, 2018.