In a typical biological system studied using computer simulations, a few large solute molecules are scattered in a bath of tiny solvent molecules. In an MD simulation of such a model, a traditional cutoff radius is dependent only on the radius of the largest solute atom present in the system. This results in a large number of 'extra' solvent interactions due to the fact that solvent molecules greatly outnumber solute molecules. However, the most useful information comes from the interactions of the solutes and we would like to limit the unnecessary calculations of solvent-solvent interactions. Since all the molecules in the system are constantly on the move, the neighbors of a given molecule are changing with time. This is normally handled using Verlet neighbor lists for computational efficiency. We need a smarter way to calculate the cutoff radius to take sizes of both of the molecules into consideration. In this thesis, I present a pair-wise cutoff method which largely solves this problem. Our new cutoff methods take into account the chemical identities of both parties of a pair of atoms. We have developed two novel cutoff methods to improve the efficiency of MD simulations. We have tested the new methods against the traditional cutoff method. In the preliminary testing, we acquired a speedup of more than three times in some model system. We have implemented the new methods into our group simulation code, OOPSE1, where another simple test showed a factor of ten speedup in some cases.