BACKGROUND: Multidrug-resistant tuberculosis programs in DOTS-Plus pilot sites in five countries. OBJECTIVES: To calculate sputum conversion time and its relationship to treatment outcome, document the frequency of culture reversions and examine concordance of smear and culture to assess the potential consequences of monitoring by smear microscopy alone. DESIGN: Retrospective cohort analysis of 1926 patients receiving individualized, second-line therapy. RESULTS: Among 1385 sputum culture-positive cases at baseline, 1146 (83%) experienced at least one culture conversion during treatment. Conversion, however, was not sustained in all patients: 201 (15%) experienced initial culture conversion and at least one subsequent culture reversion to positive; 1064 (77%) achieved sustainedculture conversion. Median time to culture conversion was 3 months. Among 206 patients whose final conversion occurred 7-18 months after the initiation of therapy, 71% were cured or had completed treatment. CONCLUSIONS: Prolonged treatment for patients with delayed conversion may be beneficial, as 71% of late converters still achieved cure or completed treatment. This has implications for programs with defined end points for treatment failure. The interval between first and final conversion among patients whose initial conversion is not sustained raises concern with respect to the ongoing debate regarding duration of treatment and the definition of cure. © 2011 The Union.