BACKGROUND: Effective tuberculosis control is compromised by a lack of clarity about the timeframe of viable Mycobacterium tuberculosis shedding after treatment initiation under programmatic conditions. This study quantifies time to conversion from smear and culture positivity to negativity in unselected tuberculosis patients receiving standardized therapy in a directly observed therapy short-course (DOTS) program. METHODS: Longitudinal cohort study following up 93 adults initiating tuberculosis therapy in Lima, Peru. Baseline culture and drug susceptibility tests (DSTs) were performed using the MBBacT, proportion, and microscopic observation drug susceptibility (MODS) methods. Smear microscopy and MODS liquid culture were performed at baseline and weekly for 4 weeks then every other week for 26 weeks. RESULTS: Median conversion time from culture positivity to culture negativity of 38.5 days was unaffected by baseline smear status. Patients with fully susceptible tuberculosis had a median time to culture conversion of 37 days; 10% remained culture positive at day 60. Delayed culture conversion was associated with multidrug resistance, regardless of DST method used; non-multidrug resistance as defined by the proportion method and MODS (but not MBBacT) was also associated with delay. Persistent day 60 smear positivity yielded positive and negative predictive values of 67% and 92%, respectively, for detecting multidrug resistance. CONCLUSIONS: Smear and culture conversion in treated tuberculosis patients takes longer than is conventionally believed, even with fully susceptible disease, and must be accounted for in tuberculosis treatment and prevention programs. Persistent day 60 smear positivity is a poor predictor of multidrug resistance. The industrialized-world convention of universal baseline DST for tuberculosis patients should become the standard of care in multidrug resistance-affected resource-limited settings.