BACKGROUND AND OBJECTIVE: In context of the End TB goal of zero tuberculosis (TB)-affected households encountering catastrophic costs due to TB by 2020, the estimation of national prevalence of catastrophic costs due to TB is a priority to inform programme design. We explore approaches to estimate the national prevalence of catastrophic costs due to TB from existing datasets as an alternative to nationally representative surveys. METHODS: We obtained, standardized and merged three patient-level datasets from existing studies on patient-incurred costs due to TB in South Africa. A deterministic cohort model was developed with the aim of estimating the national prevalence of catastrophic costs, using national data on the prevalence of TB and likelihood of loss to follow-up by income quintile and HIV status. Two approaches were tested to parameterize the model with existing cost data. First, a meta-analysis summarized study-level data by HIV status and income quintile. Second, a regression analysis of patient-level data also included employment status, education level and urbanicity. We summarized findings by type of cost and examined uncertainty around resulting estimates. RESULTS: Overall, the median prevalence of catastrophic costs for the meta-analysis and regression approaches were 11\% (interquartile range [IQR] 9-13\%) and 6\% (IQR 5-8\%), respectively. Both approaches indicated that the main burden of catastrophic costs falls on the poorest households. An individual-level regression analysis produced lower uncertainty around estimates than a study-level meta-analysis. CONCLUSIONS: This paper presents a novel application of existing data to estimate the national prevalence of catastrophic costs due to TB. This type of model could be useful for researchers and policy makers looking to inform certain policy decisions; however, some uncertainties remain due to limitations in data availability. There is an urgent need for standardized reporting of cost data and improved guidance on methods to collect income data to improve these estimates going forward.