Breast cancer is the most common female cancer worldwide. This thesis aims to evaluate the cost-effectiveness of breast cancer control across different healthcare contexts and estimate the costs of breast cancer treatment. Four case studies are presented providing detailed estimates of the cost-effectiveness of risk-based breast screening in urban China, the cost-effectiveness of population-based breast screening in rural China, the cost-effectiveness of panel genetic testing among unselected breast cancer patients in the UK and US, and cost of breast cancer treatment by stage at diagnosis in England. The economic evaluation studies on breast cancer screening show that in urban China, high-risk population-based screening for breast cancer is very likely to be cost-effective. But in rural China, breast screening among the general population reports uncertain costeffectiveness and could potentially harm women’s health due to false positives with the current screening tools. In a rural setting with such low breast cancer incidence, priority should be given to ensure that symptomatic women have proper access to diagnosis and treatment at an early stage as this will lead to mortality reductions without the usual screening harms. The economic evaluation on genetic testing based on a microsimulation model showed that unselected panel genetic-testing for all breast cancer patients is extremely costeffective compared to the current practice of family-history/clinical-criteria based genetic (BRCA)-testing for both UK and US health systems. This supports changing the current policy to expand genetic-testing to all women with breast cancer. Costs of breast cancer care increased with increasing stage of the disease at diagnosis in England. Considerable cost savings could be made if breast cancer was detected and treated earlier. Variations in breast cancer costs by age and region raise questions about the efficiency and consistency of breast cancer treatment patterns. Future research could be conducted by undertaking multiple imputation for missing data and censored-adjusted analysis.