SUMMARY BACKGROUND: There are no data comparing the 6–9 month oral three-drug Nix regimen (bedaquiline, pretomanid and linezolid [BPaL]) to conventional regimens containing bedaquiline (B, BDQ) and linezolid (L, LZD). METHODS: Six-month post end-of-treatment outcomes were compared between Nix-TB (n = 109) and 102 prospectively recruited extensively drug-resistant TB patients who received an ~18-month BDQ-based regimen (median of 8 drugs). A subset of patients received BDQ and LZD (n = 86), and a subgroup of these (n = 75) served as individually matched controls in a pairwise comparison to determine differences in regimen efficacy. RESULTS: Favourable outcomes (%) were significantly better with BPaL than with the B–L-based combination regimen (98/109, 89.9% vs. 56/86, 65.1%; adjusted relative risk ratio [aRRR] 1.35; P < 0.001) and in the matched pairwise analysis (67/75, 89.3% vs. 48/75, 64.0%; aRRR 1.39; P = 0.001), despite significantly higher baseline bacterial load and prior second-line drug exposure in the BPaL cohort. Time to culture conversion (P < 0.001), time to unfavourable outcome (P < 0.01) and time to death (P < 0.03) were significantly better or lower with BPaL than the B-L-based combinations. CONCLUSION: The BPaL regimen (and hence substitution of multiple other drugs by pretomanid and/or higher starting-dose LZD) may improve outcomes in drug-resistant TB patients with poor prognostic features. However, prospective controlled studies are required to definitively answer this question.