Background: Despite the predictive role of body weight variation in treatment outcome in multidrug-resistant tuberculosis (MDR-TB), few corroborating data are available. We studied weight variation in patients with MDR-TB to identify groups of weight change and to determine factors that influence these changes. Methods: We analyzed patients with rifampicin resistance who were treated with an MDR-TB treatment regimen between June 07, 2016 and June 22, 2018 at three major drug-resistant TB centers in Guinea. Patients were seen monthly until the end of treatment. Clinical outcome was the body mass index (BMI). We used a linear mixed model to analyze trajectories of BMI and a latent class mixed model to identify groups of BMI trajectories. Results: Of 232 patients treated for MDR-TB during the study period, 165 were analyzed. These patients had a total of 1387 visits, with a median of 5 visits (interquartile range, 3-8 visits). Monthly BMI increase was 0.24 (SE 0.02) per kg/m2. Factors associated with faster BMI progression were success of MDR-TB treatment (0.24 [SE 0.09] per kg/m2; p = 0.0205) and absence of lung cavities on X-ray (0.18 [0.06] per kg/m2; p = 0.0068). Two groups of BMI change were identified: rapid BMI increase (n = 121; 85%) and slow BMI increase (n = 22; 15%). Patients in the slow BMI increase group were mostly female (68%) had no history of TB treatment (41%), had a positive HIV infection (59%), and had a more severe clinical condition at baseline, characterized by a higher frequency of symptoms including depression (18%), dyspnea (68%), poor adherence to MDR-TB treatment (64%), lower platelet count, and higher SGOT. These patients also had a longer time to initial culture conversion (log-rank test: p = 0.0218). Conclusion: Quantitative BMI data on patients with MDR-TB treated with a short regimen allowed the identification of subgroups of patients with different trajectories of BMI and emphasized the usefulness of BMI as a biomarker for the monitoring of MDR-TB treatment outcome