The standard theoretical description used to describe electron transfer is Marcus theory, which maps the polarization of the solvent surrounding the reactants onto a reaction coordinate, q. The questions we address in this paper are: How many and what types of solvent degrees of freedom constitute q? Is it appropriate to treat the solvent as a dielectric continuum? Our approach to answer these questions relies on the study of the simplest possible charge transfer systems: we choose atomic systems that have only electronic degrees of freedom so that any spectroscopic changes that occur during the course of the reaction directly reflect the motions of the surrounding solvent. Our methods for characterizing these systems consist of both molecular dynamics (MID) simulations and femtosecond pump-probe experiments, Using MID, we rind that even though solvent rotational motions appear to dominate the electronic relaxation when only the solute's charge changes, the slow translational motions of the few closest solvent molecules control the solvation dynamics when realistic reactant size changes are taken into account. Moreover, we see that the linear response approximation, an assumption inherent in the use of dielectric continuum theories, fails when reactant size changes and solvent translational motions are involved. Our experimental approach focuses on the study of the charge-transfer-to-solvent (CTTS) transition of the sodium anion (Na-). We find that the charge-transfer rate of photoexcited sodide in tetrahydrofuran is similar to3 times slower than what would be expected by assuming that dielectric solvation was the dominant driving force for electron transfer. This suggests that the slow solvent translational motions needed to accommodate the reactant size change are rate-limiting for the charge transfer process, consistent with the simulations. The electron appearance and recombination kinetics also show that even though the charge transfer rate is roughly independent of excitation energy, the distance over which the electron is ejected depends sensitively on the excitation energy. Moreover, the detached electrons recombine with their Na atom partners to regenerate the parent sodide ions on two vastly different time scales. The best way to explain the electron recombination dynamics invokes the existence of two kinds of solvated electron: geminate sodium atom contact pairs. Our molecular picture of the charge-transfer process is that low-energy excitation produces a CTTS excited-state wave function confined within the original anion solvent cavity, leading to production of a sodium atom:solvated electron contact pair that can recombine in about one picosecond. The use of high excitation energies produces CTTS excited-state wave functions with greater curvature and spatial extent, allowing the electron to localize further from the parent in a long-lived (greater than or equal to 200 ps) solvent-separated contact pair. or to be ejected into the solvent. Independent of the excitation energy, it is the relatively slow translational motions of First-shell solvent molecules that are responsible for the electron detachment. All the results are compared to previous experimental and theoretical work.