Abstract: The female breast is one of the most radiosensitive organs, and mean glandular tissue (MGD) is regarded as the most reasonable dose descriptor related to the risk of radiation-induced breast cancer. To estimate the MGD more precisely, a mathematical model with detailed structures of breast was constructed, which includes skin, subcutaneous fat layer, retromammary fat layer, Cooper’s ligaments, intraglandular fats, ductal trees, lobules, lactiferous sinus, nipple and so on. The mathematical model was then voxelized to a voxel model. The voxel breast model was compressed in the cranio caudal direction for mammography. The compressed breast model was combined with the Chinese reference adult female voxel model (CRAF), and Monte Carlo simulation of mammography was performed based on it with Geant4. A series of glandular tissue dose conversion factors for mammography were calculated. The data calculated in this work are smaller than the data from specification for testing of quality control in mammography of China. However, the difference between the data calculated in this work and those from American College of Radiology (ACR) is small.