The problem of photoinduced donor-acceptor electron transfer in liquid solution is analyzed to obtain an understanding of the relationship between approximate treatments of the role of diffusion in electron transfer, that is, the Collins-Kimball approach, and a detailed analysis of the problem. It is shown why previous analyses of experimental data have yielded distance dependences of electron transfer that are much too long range. From an appropriate fitting of the nonstationary kinetics of donor fluorescence quenching by diffusion-assisted electron transfer, the effective radii and the steady-state constants associated with electron transfer are found for a donor-acceptor system studied experimentally in seven solvents with different viscosities. The dependence of diffusion agrees with the one predicted theoretically for electron transfer having a distance-dependent transfer rate initially taken to be exponential with distance. In the fast-diffusion limit, the dependence on the rate of diffusion is well approximated by the Collins-Kimball relationship, which permits the kinetic rate constant and the effective radius associated with diffusion-induced quenching to be extracted from the experimental data. The effective radius is then related to the electron transfer rate with arbitrary distance dependence. From this relationship, the tunnelling length for both exponential and Marcus-type rates is obtained from the data analysis, and it is demonstrated that the latter is almost twice as long as the former. For the Marcus transfer rate, it is found that the Marcus parameter beta = 1.2 Angstrom(-1) (beta = 2/tunnelling length), which is in accord with previous measurements on a variety of systems. The theoretical analysis presented here resolves the apparent discrepancies between early measurements of very long tunnelling lengths in liquid systems and physically reasonable values of beta approximate to 1 Angstrom(-1).