OBJECTIVES: Dynamic transmission models used in cost-effectiveness analyses of Tuberculosis (TB) interventions often apply various assumptions regarding disease progression and intervention implementation. The influence of these varying assumptions on the model’s cost and health outcome projection was explored in this study. METHODS: One base case TB dynamic transmission model and eight model variants were developed. The base case model was built based on the common assumptions of previously published models. In each model variant, one of these assumptions was modified. All models were utilized to project the cost and health outcome of implementing Xpert.MTB/RIF (Xpert) in South Africa from the healthcare system perspective. The same input parameters, taken from a previous study, were used for all models. Population specific parameters, i.e. entry, contact, and detection rate, were obtained by calibrating the models to the number of new and diagnosed cases derived from a WHO report. The model variants’ cost and health outcome projection, reported as the Incremental Cost Effectiveness Ratio (ICER) in US dollars per DALY averted, was compared to the base case model’s ICER. RESULTS: The base case model projected that Xpert implementation would be cost-saving, with an ICER of -$38.64/DALY averted. One model variant which assumed two levels of latent disease estimated a high contact rate. The high contact rate caused a high force of infection which limited Xpert’s performance and resulted in a high cost for Xpert implementation and an ICER of $86.52/DALY averted. A similar projection, with an ICER of $16.01/DALY averted, was found in another model variant which assumed TB cases under treatment to remain partially infectious. CONCLUSIONS: Assumptions could substantially influence a model’s cost and health outcome projection, which could affect the decision on intervention implementation. This finding highlights the importance of reporting assumptions transparently and performing adequate uncertainty analyses, including structural uncertainty analysis.