Efforts to compare thermodynamic predictions of simulations with experiment have been thwarted by the approximations used in modeling the long-range electrostatic interactions. In a recent article Hummer et al. proposed an explicit formula that accounts for most of the finite-size effect in simulations of ion charging using periodic boundary conditions. In this Letter we have verified the numerical results of Hummer et al. and we provide a simple theoretical argument that explains why the solvent polarization free energy contains a contribution that largely cancels the lattice self-energy. The cancellation depends on the fact that the solvent is a high dielectric. We propose a finite size correction formula that can be applied in both the low and high dielectric limits.