Contrast-enhanced dual energy mammography (CEDEM) has been shown effective in enhancing the visibility of breast cancer in diagnostic mammographic images. The simulation and phantom studies of CEDEM were implemented to enhance acquisition techniques of a dedicated cone-beam breast CT scanner fabricated in our laboratory. CEDEM images were computer-simulated numerically, and the results were compared and validated with physical experiments using a custom-made breast-equivalent phantom with pseudo anatomical structures. The technique factors to be optimized included low- and high-energy x-ray spectra (kVp), spectra filters and mAs (x-ray tube current × scan time in second). Signal-difference-to-noise ratio (SDNR) of the log-weighted subtracted image was computed to optimize subtraction parameters using several combinations of breast compositions and thicknesses. Both simulation and phantom studies suggest that 40 kVp (filtered with 1.5 mm Al) and 50 kVp (filtered with 1.5 mm Al and 0.3 mm Cu) were the optimal spectra with the highest SDNR on the dual energy subtracted image of a 6-cm, 50% glandular breast. Breast composition and thickness were found to have little influence on the optimized technique factors. We conclude that the simulation and phantom studies demonstrated the effectiveness of CEDEM using the prototype breast CT scanner in producing high-quality, lesion-enhanced images.