In vivo counting measurements, used for the monitoring of workers with internal contamination risks, are based on the use of calibration physical phantoms. However, such phantoms do not exist for female subjects. Computational calibration using numerical representations, Mesh and non-uniform rational basis spline (NURBS) geometries, was thus considered. The study presented here is focused on the creation of different female thoracic phantoms with various breast sizes and chest girths. These 3D models are used to estimate the radiation attenuation with morphology and the resulting variation of the calibration coefficient of a typical 4-germanium in vivo counting system. A basic Mesh female thoracic phantom was created from the International Commission on Radiological Protection Adult Female Reference Computational Phantom. Using this basic phantom, different chest girths (85, 90, 100, 110, and 120) and cup sizes (A to F) were created representing the most common thoracic female morphologies, as recommended by the available and relevant literature. Variation of breast tissue composition and internal organ volumes with morphology were also considered. As a result, 34 thoracic female phantoms were created combining different cup sizes and chest girths. For the 85 chest girth, at very low energies (15 keV), a relative counting efficiency variation of about 85% was observed between the A and E cups. As a result of this study, breast size dependent calibration coefficients, between 15 keV and 1.4 MeV, were obtained and tabulated for a typical lung counting germanium system.