OBJECTIVE: To estimate the cost-effectiveness of directly observed treatment compared to conventional therapy in reducing the spread of multidrug-resistant tuberculosis, for an industrialised country (represented by the United States of America) and a developing country (South Africa). METHODS: Monte Carlo analysis using published data on probability, cost and health impact. RESULTS: In both countries, directly observed treatment is the dominant strategy, yielding cost savings and improved health outcomes. Cost savings for directly observed treatment relative to conventional therapy become more significant as more expensive second-line drugs are used in treatments. CONCLUSIONS: The cost-effectiveness of directly observed treatment relative to conventional therapy is demonstrated for both the USA and South Africa. Cost savings are more pronounced (especially for South Africa) as the likelihood of multidrug-resistant tuberculosis increases and more expensive second-line therapies are used. Given that health care resources are more severely constrained in developing countries, the data contained in this study are useful in guiding the design of policies for the effective management of multidrug-resistant tuberculosis in settings with limited resources.