Background: Even with newer and repurposed anti-TB drugs almost a third of patients with XDR-TB have unfavourable outcomes. In patients with localised disease and adequate pulmonary reserve, surgery is an important adjunctive treatment. However, there are no outcome data from TB endemic countries, and the prognostic significance of pre-operative PET-CT findings remains unknown. Objectives: To report outcomes for resectional surgery in our setting, and to study whether PET activity outside of the resection influences treatment outcomes. Methods: A retrospective study of all XDR-TB patients undergoing surgery at Groote Schuur Hospital (GSH) between July 2010 and December 2016 was performed. PET-CT was performed in a subgroup. Patients were followed up to determine treatment outcomes at 24-months post- surgery. Treatment success and failure, including all-cause mortality, was determined. Results: In total, 35 patients underwent surgery. The mean age was 36, 49% were male and 26% were HIV-infected. Pneumonectomy was the most common procedure (57%). Three patients (9%) were lost to follow up by 24 months. Total all-cause mortality was 34%. Treatment success was achieved in 15/35 (43%). In patients who underwent pre-operative PET-CT, there were no overall radiological features or PET parameters that were found to be prognostic for treatment failure. Conclusion: Resectional surgery for DR-TB in combination with chemotherapy resulted in cure in less than half of patients. Our data do not support the use of PET-CT to preselect patients or prognosticate about their outcome. These data inform clinical practice and underscore the need to support antibiotic stewardship strategies in TB-endemic settings.