There are limited data on combining delamanid and bedaquiline in drug-resistant tuberculosis (DR-TB) regimens. Prospective long-term outcome data, including in HIV-infected persons, are unavailable.We prospectively followed up 122 South African patients (52.5% HIV-infected) with DR-TB and poor prognostic features between 2014 and 2018. We examined outcomes and safety in those who received a bedaquiline-based regimen (n=82) compared to those who received a bedaquiline–delamanid combination regimen (n=40).There was no significant difference in 6-month culture conversion (92.5% versus 81.8%; p=0.26) and 18-month favourable outcome rate (63.4% versus 67.5%; p=0.66) in the bedaquiline versus the bedaquiline–delamanid combination group, despite the latter having more advanced drug resistance (3.7% versus 22.5% resistant to at least five drugs; p=0.001) and higher pre-treatment failure rates (12.2% versus 52.5% with pre-treatment multidrug-resistant TB therapy failure; p<0.001). Although the proportion of prolongation of the QT interval corrected using Fridericia's formula was higher in the combination group (>60 ms from baseline (p=0.001) or >450 ms during treatment (p=0.001)), there were no symptomatic cases or drug withdrawals in either group. Results were similar in HIV-infected patients.A bedaquiline–delamanid combination regimen showed comparable long-term safety compared to a bedaquiline-based regimen in patients with DR-TB, irrespective of HIV status. These data inform regimen selection in patients with DR-TB from TB-endemic settings.