Patterns of the shifts in bridging cyanide-stretching frequencies have been examined in several fully saturated, mu-cyano, bi- or trimetallic transition-metal donor-acceptor (D/A) complexes. An earlier (Watzky, M. A.; et al. Inorg. Chem. 1996, 35, 3463) inference that the bridging ligand nuclear and the D/A electronic coordinates are entangled is unequivocally demonstrated by the 32 cm(-1) lower frequency of v(CN) for (NH3)(5)Cr(CNRu(NH3)(5))(4+) than for the cyanopentaamminechromium(III) parent. This contrasts to the 41 cm(-1) increase in v(CN) upon ruthenation of (NH3)(5)RhCN2+. More complex behavior has been found for cis and trans trimetallic, donor-acceptor complexes. The symmetric combination of CN- stretching frequencies in trans-Cr-III(MCL)(CNRuII(NH3)(5))(2)(5+) complexes (MCL = a tetraazamacrocyclic ligand) shifts 100-140 cm(-1) to lower frequency, and the antisymmetric combination shifts less than about 30 cm(-1). This contrast in the shifts of the symmetric and the antisymmetric combinations of the CN stretches persists even in a trans complex with no center of symmetry. Two CN stretches have also been resolved in an analogous cis complex, and both shift to lower frequency by about 60 cm(-1). The net shift, summed over all the CN-stretching frequencies, is about the same for the bis-ruthenates of related dicyano complexes. A simple, symmetry-adapted perturbation theory treatment of the coupled vibrations is employed to deal with the opposing effects of the "kinematic" shifts (delta) of v(CN) to higher frequency, expected in the absence of D/A coupling, and shifts (f) of v(CN) to lower frequency that occur when D/A coupling is large. The Rh(III)and Cr(III)-centered complexes correspond to different limits of this model: delta > f and delta < f, respectively. When referenced by means of this model to complexes with Rh(III) acceptors, the shifts in trimetallic complexes, summed over the symmetric and antisymmetric combinations of CN stretches, are about twice those of bimetallic complexes. Similarly referenced and summed over all bridging CN frequencies, the shifts of v(CN) to lower energies are proportional to the oscillator strength of the electronic, donor-acceptor charge-transfer transition. The simplest interpretation of this correlation is that the donor-acceptor coupling in these systems is a function of the nuclear coordinates of the bridging ligand. This behavior of these complexes is semiquantitatively consistent with expectation for CN--mediated vibronic (pseudo-Jahn-Teller) coupling of neighboring donors and accepters, and the observed Ru-II/CN- CT absorption parameters can be used in a simple, semiclassical vibronic model to predict shifts in v(CN) that are in reasonable agreement with those observed.