The last few decades have seen enormous progress in both geometric subdivision, and physics-based simulation techniques. Mesh-based dynamic systems often require both subdivision and physical simulation for realistic and accurate results. However, the simulation parameters have been independent of the subdivision parameters, and vice-versa. This paper attempts to bridge this gap. We propose a hybrid approach that combines the physics-based simulation techniques and geometric subdivision algorithms, and demonstrate a mass-spring based system with physics-based butterfly subdivision. The initial subdivision coefficients are extracted using the physical properties of the base (L_0) mesh. Latter subdivision steps generate both the geometric and physical properties of the subdivided (L_k) mesh. This approach conserves mass, center of gravity, linear momentum and external force, and minimizes the distance between the L_k and L_k+1 meshes, at any time step. Our approach is general, efficient, and will serve as a foundation for many applications in many fields.