Research on the microstructure-based modeling of composite dielectrics for capacitor applications is described. Methods for predicting the composite dielectric permittivity and internal electric field distributions within the microstructure using finite difference quasi-electrostatic modeling are described, along with methods of generating realistic model spaces of particulate microstructures. An existing algorithm for generating random, monosized spheres-in-a-dielectric matrix model spaces is modified to allow the treatment of bimodal composites in which small particles are deliberately segregated into the spaces between large particles. Such composites can have substantially higher total volumetric filling fractions of particles, leading to higher composite permittivity. The variations in permittivity with the filling fractions of bimodal inclusions are studied with the new model, with cases covering three different types of polymer matrix material. The effect of the small particle additions on the electric field statistics within the polymer matrix is also explored.