In "outer sphere" electron transfer reactions, motions of the solvent molecules surrounding the donor and acceptor govern the dynamics of charge flow. Are the relevant solvent motions determined simply by bulk solvent properties such as dielectric constant or viscosity? Or are molecular details, such as the local solvent structure around the donor and acceptor, necessary to understand how solvent motions control charge transfer? In this paper, we address these questions by using ultrafast spectroscopy to study a photoinduced electron transfer reaction with only electronic degrees of freedom: the charge-transfer-to-solvent (CTTS) reaction of Na- (sodide). Photoexcitation of Na- places the excited CTTS electron into a solvent-bound excited state; motions of the surrounding solvent molecules in response to this excitation ultimately lead to detachment of the electron. The detached electron can then localize either in an "immediate" contact pair (in the same cavity as the Na atom), which undergoes back electron transfer to regenerate Na- in similar to1 ps, or in a "solvent-separated" contact pair (one solvent shell away from the Na atom), which undergoes back electron transfer in tens to hundreds of picoseconds. We present detailed results for the dynamics of each step of this reaction in several solvents: the ethers tetrahydrofuran, diethyl ether and tetrahydropyran and the amine solvent hexamethylphosphoramide (HMPA). The results are interpreted in terms of a kinetic model that both incorporates spectral shifting of the reaction intermediates due to solvation dynamics and accounts for anisotropic spectral diffusion in polarized transient hole-burning experiments. We find that the rate of CTTS detachment does not correlate simply with any bulk solvent properties, but instead appears to depend on the details of how the solvent packs around the solute. In contrast, the rate for back electron transfer of solvent-separated contact pairs varies inversely with solvent polarity, indicating a barrier to recombination and suggesting that this reaction lies in the Marcus inverted regime. For immediate contact pairs, the rate of recombination varies directly with solvent polarity in the ethers but is slowest in the highly polar solvent HMPA, suggesting that the spatial extent of the solvated electron in each solvent is one of the major factors determining the recombination dynamics. The fact that each step in the reaction varies with solvent in a different way implies that there is not a single set of solvent motions or spectral density that can be used to model all aspects of electron transfer. In addition, all of the results and conclusions in this paper are compared in detail to related work on this system by Ruhman and co-workers; in particular, we assign a fast decay seen in the near-IR to solvation of the CTTS p-to-p excited-state absorption, and polarization differences observed at visible probe wavelengths to anisotropic bleaching of the Na- CTTS ground state. (C) 2003 American Institute of Physics.