INTRODUCTION: Identification of good prognostic marker for tuberculosis (TB) treatment response is a necessary step on the path towards a surrogate marker to reduce TB trial duration.METHODS: We performed a retrospective analysis on routinely collected data in 6 drug-resistant TB (DRTB) programs. Culture conversion, defined as two consecutive negative cultures, was assessed, and performance of culture conversion at Month 2 and Month 6 to predict treatment success were explored. To explore factors associated with positive predicted value (PPV) and the specificity of culture conversion, a multinomial logistic regression was fitted.RESULTS: This study included 634 patients: 68.5% were males; the median age was 35 years, 75.2% were previously treated for TB, 59.4% were resistant only to isoniazid and rifampicin and 18.1% resistant to fluoroquinolones. Culture conversion at Month 2 and 6 showed similar PPV while specificity was much higher for culture conversion at Month 2: 91.3% (95%CI 86.1-95.1). PPV of culture conversion at Month 2 did not vary strongly according to patients' characteristics, while specificity was slightly higher among patients with fluoroquinolone-resistant strains.CONCLUSION: Culture conversion at Month 2 is an acceptable prognostic marker for MDR-TB treatment. Considering the advantage of using an earlier marker, further evaluation as a surrogate marker is warranted to shorten TB trials.