The azide (pi(nb))(sigma) --> Cu charge transfer (CT) spectra of two Cu(II) mu-1,1-azido dimers, [Cu-2(tmen)(2)(N-3)(OH)]-(ClO4)(2) (tmen = tetramethylethylenediamine) and [Cu-2(tbupy)(4)(N-3)(2)](ClO4)(2) (tbupy = tert-butylpyridine), are recorded by solution and single-crystal polarized absorption spectroscopy. The spectra are interpreted with a valence bond configuration interaction (VBCI) model. This model parametrizes antiferromagnetic and ferromagnetic interactions present in the manifold of CT states in terms of transfer integrals between the metal d orbitals and the highest occupied MO of azide, (pi(nb))(sigma), and the lowest unoccupied orbital of azide, (pi*)(sigma). These two transfer integrals are found to be approximately equal, leading to a compensation of ferro- and antiferromagnetic interactions in the CT state, which in turn is the reason for the observed ferromagnetic coupling in the ground state. The origin of the ferromagnetic interaction is a polarization of the unpaired metal electrons by an unpaired electron in the azide (pi*)(sigma) orbital in a ligand excited (LE) state. The spin polarization model (Charlot, M.-F.; Kahn, O.; Chaillet, M.; Larrieu, C. J. Am. Chem. Soc. 1986, 108, 2574) is found to be a spin-Hamiltonian description of this ferromagnetic interaction through the azide pi* orbital. In terms of a molecular orbital description, the combined interaction of azide (pi(nb))(sigma) and (pi*)(sigma) orbitals drive the mu-1,1 systems toward accidental degeneracy between HOMO and LUMO, the condition for ferromagnetic coupling in the "active electron approximation".