© 2019 Elsevier Ltd Background: Delamanid is one of two recently approved drugs for the treatment of multidrug-resistant tuberculosis. We aimed to evaluate the safety and efficacy of delamanid in the first 6 months of treatment. Methods: This randomised, double-blind, placebo-controlled, phase 3 trial was done at 17 sites in seven countries (Estonia, Latvia, Lithuania, Moldova, Peru, the Philippines, and South Africa). We enrolled eligible adults (>18 years) with pulmonary multidrug-resistant tuberculosis to receive, in combination with an optimised background regimen developed according to WHO and national guidelines, either oral delamanid (100 mg twice daily) for 2 months followed by 200 mg once daily for 4 months or placebo (same regimen). Patients were centrally randomised (2:1) and stratified by risk category for delayed sputum culture conversion. Primary outcomes were the time to sputum culture conversion over 6 months and the difference in the distribution of time to sputum culture conversion over 6 months between the two groups, as assessed in the modified intention-to-treat population. The trial is registered at ClinicalTrials.gov, number NCT01424670. Findings: Between Sept 2, 2011, and Nov 27, 2013, we screened 714 patients, of whom 511 were randomly assigned (341 to delamanid plus optimised background regimen [delamanid group] and 170 to placebo plus optimised background regimen [placebo group]) and formed the safety analysis population. 327 patients were culture-positive for multidrug-resistant tuberculosis at baseline and comprised the efficacy analysis population (226 in the delamanid group and 101 in the placebo group). Median time to sputum culture conversion did not differ between the two groups (p=0·0562; modified Peto-Peto), with 51 days (IQR 29–98) in the delamanid group and 57 days (43–85) in the placebo group; the hazard ratio was 1·17 (95% CI 0·91–1·51, p=0·2157). 501 (98·0%) of 511 patients had at least one treatment-emergent adverse event. 136 (26·6%) of 511 patients had at least one serious treatment-emergent adverse event; the incidence was similar between treatment groups (89 [26·1%] of 341 patients for delamanid and 47 [27·6%] of 170 for placebo). Deaths related to treatment-emergent adverse events were similar between groups (15 [4·4%] of 341 for delamanid and six [3·5%] of 170 for placebo). No deaths were considered to be related to delamanid. Interpretation: The reduction in median time to sputum culture conversion over 6 months was not significant in the primary analysis. Delamanid was well tolerated with a highly characterised safety profile. Further evaluation of delamanid is needed to determine its role in a rapidly evolving standard of care. Funding: Otsuka Pharmaceutical.