An anatomically accurate model of the forearm and hand musculo-skeletal system using finite element geometries is presented. Anatomical data has been digitized from the male Visible Human dataset to create meshes which accurately approximate each bone and muscle volume. Each muscle's anatomical structure has been accounted for via the topology of the initial mesh generation. Cylindrical muscles have been modeled using collapsed bicubic-linear elements, while more complex muscle topologies have required tricubic elements. Bifurcating muscle meshes combine 1D elements for the tendons and 3D elements for the adjoining muscle. The fitting process of each mesh to its dataset was carried out using a least-squares algorithm, to minimize the distances between the mesh and the data points. Sobelov smoothing constraints have been implemented to account for sparse and scattered data. The fitted forearm muscles contain 1085 nodes, 797 elements, and an average RMS error of 1.8207 mm; while the fitted hand muscles use 509 nodes, 274 elements, and an average RMS error of 1.1136 mm. Applications of the model as a framework for visualization of anatomical data relevant for biomedical and medical education are discussed. With mesh customization and further functional modeling this provides the basis for surgical training and functional development.