In recent years, a number of quantum mechanical-molecular mechanical (QM/MM) enzyme studies have investigated the dependence of reaction energetics on the size of the QM region using energy and free energy calculations. In this study, we revisit the question of QM region size dependence in QM/MM simulations within the context of energy and free energy calculations using a proton transfer in a DNA base pair as a test case. In the simulations, the QM region was treated with a dispersion-corrected AM1/d-PhoT Hamiltonian, which was developed to accurately describe phosphoryl and proton transfer reactions, in conjunction with an electrostatic embedding scheme using the particle-mesh Ewald summation method. With this rigorous QM/MM potential, we performed rather extensive QM/MM sampling, and found that the free energy reaction profiles converge rapidly with respect to the QM region size within ca. ±1 kcal/mol. This finding suggests that the strategy of QM/MM simulations with reasonably sized and selected QM regions, which has been employed for over four decades, is a valid approach for modeling complex biomolecular systems. We point to possible causes for the sensitivity of the energy and free energy calculations to the size of the QM region, and potential implications.