OBJECTIVE: To investigate whether diabetes mellitus (DM) influences TB treatment outcomes.METHODS: This was a retrospective observational cohort study of all notified TB cases from a large London TB centre over a 5-year period. WHO criteria were used to define TB treatment outcomes.RESULTS: The prevalence of DM at TB treatment initiation was 15% (126/838). Most patients (83.3%, 105/126) were on hypoglycaemic treatment and well-controlled (median glycated haemoglobin 53.5 mmol/mol). DM patients were older, more likely to be of Asian ethnicity and had a higher pre-treatment weight. Time from presentation to treatment initiation was longer (median 87.5 vs. 63 days; P < 0.001), while they were significantly more comorbid (median Charlson Comorbidity Index 3 vs. 0; P < 0.001). Overall, favourable treatment outcomes were recorded for 89.5% of patients (87.7% vs. 89.8% for DM and non-DM patients respectively, P = 0.52). In multivariable analysis, DM was not associated with unfavourable TB treatment outcomes (OR 0.49, 95% CI 0.23-1.04, P = 0.06). Independent predictors of unfavourable outcome included age, cavitation, chronic neurological disease and malignant neoplasm.CONCLUSIONS: In a well-resourced setting, with predominantly well-controlled DM patients on treatment, DM was not an independent predictor of unfavourable TB treatment outcomes.