The finite-element approach of absolute nodal coordinate formulation (ANCF) is a possible way to simulate the deployment dynamics of a large-scale mesh reflector of satellite antenna. However, the large number of finite elements of ANCF significantly increases the dimension of the dynamic equations for the deployable mesh reflector and leads to a great challenge for the efficient dynamic simulation. A new parallel computation methodology is proposed to solve the differential algebraic equations for the mesh reflector multibody system. The mesh reflector system is first decomposed into several independent subsystems by cutting its joints or finite-element grids. Then, the Schur complement method is used to eliminate the internal generalized coordinates of each subsystem and the Lagrange multipliers for joint constraint equations associated with the internal variables. With an increase of the number of subsystems, the dimension of simultaneous linear equations generated in the numerical solution process will inevitably increase. By using the multilevel decomposition approach, the dimension of the simultaneous linear equations is further reduced. Two numerical examples are used to validate the efficiency and accuracy of the proposed parallel computation methodology. Finally, the dynamic simulation for a 500 s deployment process of a complex AstroMesh reflector with over 190,000 generalized coordinates is efficiently completed within 78 hrs.