The aim of this study is to investigate the feasibility of water and lipid as calibration phantoms for accurate dual energy breast density quantification. Dual energy calibration was performed on a mammography system based on scanning multi-slit Si strip photon-counting detectors using plastic water and adipose-equivalent phantoms as the basis materials. Two different methods were used to convert the dual energy decomposition measurements in plastic phantom thicknesses into the true water and lipid basis materials. The first method was based entirely on the theoretically calculated effective attenuation coefficients of the investigated materials in the mammographic energy range. The conversion matrix was determined through the linear least-squares fitting of the target material using the calculated effective attenuation coefficients of water and lipid. The second method was based on experimental calibration with plastic water phantom, adipose-equivalent phantom, and its correlation to known water and lipid thicknesses. These two methods were then validated by using an independent measurement of water and lipid mixture phantoms and postmortem breasts. The correlation between the dual energy decomposition measurements and the known values was evaluated using linear regression analysis. The averaged root-mean-square errors for water density quantification derived from the theoretical and experimental conversions were 8.6% and 1.6%, respectively. The postmortem breast tissue study also indicates that the experimentally acquired conversion coefficient improved the accuracy in water density quantification, in comparison with that from the theoretical conversion. The results show that conversion of the dual energy measurements into water and lipid thicknesses improves the accuracy in breast tissue decomposition.