Polynuclear transition-metal complexes, such as Fe-S clusters, are the prosthetic groups in a large number of metalloproteins and serve as temporary electron storage units in a number of important redox-based biological processes. Polynuclearity distinguishes clusters from mononuclear centers and confers upon them unique properties, such as spin ordering and the presence of thermally accessible excited spin states in clusters with paramagnetic sites, and fractional valencies in clusters of the mixed-valence type. In an earlier study we presented an effective-mode (EM) analysis of electron transfer from a binuclear mixed-valence donor with paramagnetic sites to a mononuclear acceptor which revealed that the cluster-specific attributes have an important impact on the kinetics of long-range electron transfer. In the present study, the validity of these results is tested in the framework of more detailed theories which we have termed the multimode semiclassical (SC) model and the quantum-mechanical (QM) model. It is found that the qualitative trends in the rate constant are the same in all treatments and that the semiclassical models provide a good approximation of the more rigorous quantum-mechanical description of electron transfer under physiologically relevant conditions. In particular, the present results corroborate the importance of electron transfer via excited spin states in reactions with a low driving force and justify the use of semiclassical theory in cases in which the QM model is computationally too demanding. We consider cases in which either one or two donor sites of a dimer are electronically coupled to the acceptor. In the case of multiconnectivity, the rate constant for electron transfer from a valence-delocalized (class-III) donor is nonadditive with respect to transfer from individual metal sites of the donor and undergoes an order-of-magnitude change by reversing the sign of the intradimer metal-metal resonance parameter (beta). In the case of single connectivity, the rate constant for electron transfer from a valence-localized (class-II) donor can readily be tuned over several orders of magnitude by introducing differences in the electronic potentials at the two metal sites of the donor. These results indicate that theories of cluster-based electron transfer, in order to be realistic, need to consider both intrinsic electronic structure and extrinsic interactions of the cluster with the protein environment.