ABSTRACT The impact of heteroresistance on tuberculosis (TB) treatment outcomes is unclear, as is the role of different rifampin and isoniazid exposures on developing resistance mutations. Hollow fiber system model of TB (HFS-TB) units were inoculated with drug-susceptible Mycobacterium tuberculosis ( Mtb ) and treated with isoniazid and rifampin exposure identified in a clinical trial as leading to treatment failure and acquired drug resistance. Systems were sampled for drug concentration measurements, estimation of total and drug-resistant Mtb , and small molecule overlapping reads (SMOR) analysis for the detection of heteroresistance. In the second HFS-TB study, systems were inoculated with an isoniazid-resistant clinical strain and treated with various combinations of isoniazid, rifampin, moxifloxacin, and levofloxacin for 28 days. Linear regression and exponential decline models were used for data analysis. Suboptimal isoniazid and rifampin exposures failed to kill drug-susceptible Mtb in the HFS-TB. Standard susceptibility methods failed to detect drug resistance, but SMOR detected isoniazid and rifampin heteroresistance, as well as fluoroquinolone, to which bacilli were not exposed. rpoB mutations arising from low rifampin exposures were Q513K and H526N, whereas those from regimen adequate rifampin but low isoniazid concentrations were S531L. Moxifloxacin-rifampin combination sterilized the HFS-TB units inoculated with the isoniazid-resistant Mtb in 14 days compared with 21 days of treatment with levofloxacin-rifampin, with no further emergence of drug resistance. Early detection of isoniazid and rifampin heteroresistance could provide an opportunity to individualize the therapy and protect fluoroquinolones when added to the MDR-TB treatment regimen.