MRI breast images with microwave tomography is combined through a soft prior technique, incorporating spatial information from MRI into the regularization process of the image reconstruction algorithm. Numerical experiments were completed on MR-derived 3D breast geometries with different parenchymal densities, as well as a simulated tumor. When the soft prior technique was applied, relative root mean square error values decreased in all cases across all breast densities. The incorporation of structural priors significantly increased the recovered contrast between tumor and fibroglandular tissue. This study confirmed that the soft prior regularization algorithm is robust in our 3D microwave image reconstruction.