Fracture healing and growth of the bones are highly related to the stress level. Numerical analysis of stresses is the most effective means to determine the stress level, but it usually requires sufficient resolution to ensure an accurate description of geometry features of bones. In this paper, high-fidelity smoothed finite element method (S-FEM) skull models are created using computed tomography (CT) and micro-computed tomography (μCT) images of a juvenile pig skull. The material properties of the heterogeneous bone are modeled by a varying distribution of Young’s modulus mapped to each element and smoothing domain to accurately capture the high heterogeneity. Different types of S-FEM models, including node-based, edge-based, and facebased, are developed for this high-fidelity modeling work. It is found that S-FEM has higher accuracy, in terms of displacements, stresses, and strain energy, compared to the traditional finite element method (FEM).