Setting: National tuberculosis control programme, Vanuatu. Objective: To assess tuberculosis (TB) trends, characterise sputum smear-positive patients with non-conversion at 2 months and assess their treatment outcomes. Design: Evaluation of programme data over a 9-year period (2004–2012), comparing 2-month sputum non-converters (delayed converters) with sputum smear converters diagnosed in 2011 and 2012. Results: Annual TB case numbers were similar over the study period, with an average TB notification rate of 58 per 100 000 population. Of 417 sputum smear-positive cases, 74 (18%) were delayed converters. Delayed converters were more likely than converters (88% vs. 79%) to have had high pre-treatment sputum smear grades (OR 2.5, 95%CI 0.97–6.45). Among delayed converters, treatment adherence was high (99% good adherence), outcomes were generally good (90% treatment success, 85% cure, 4% treatment failure) and no drug resistance was detected. Deaths were unexpectedly common among converters (11/80, 14%), with significantly more deaths in Tafea than in Shefa Province (7/58 vs. 2/80, OR 5.35, 95%CI 1.07–26.79). Tafea Province also had the greatest number of delayed converters (30/74, 40.5%) and the highest TB incidence rate. Conclusion: Delayed sputum conversion was relatively uncommon, and was not associated with adverse outcomes or drug resistance. Regional differences require further investigation to better understand local factors that may compromise patient management.