This paper reports a detailed theoretical study of the interaction between a central low-spin d(8) nickel ion and two N,N-coordinating diiminobenzosemiquinonate(1-) ligands in a square planar arrangement. Such complexes have recently attracted much attention due to their unusual bonding patterns, structures, optical, and magnetic properties. Geometry optimizations using various levels of density functional theory (DFT) result in excellent agreement with the experimentally determined structure and in particular reproduce the quinoidal distortions in the aromatic rings well. A detailed analysis of the orbital structure reveals that the complex features essentially two strongly interacting ligand radicals which interact with each other via an efficient superexchange mechanism that is mediated by a back-bonding interaction to the central metal. An analysis of the broken symmetry DFT wave function is presented and a new index for the diradical character is proposed which shows that [Ni(L-ISQ)(2)] has a diradical character of similar to77%. These results are in full agreement with elaborate multireference post-Hartree-Fock ab initio calculations for [Ni(L-ISQ)(2)] using the difference dedicated configuration interaction (DDCI) method as well as second-order multireference Moller-Plesset (MR-MP2) theory, which give diradical characters of 65-80%. On the basis of these calculations our best estimate for the singlet-triplet gap in this system is 3096 cm(-1). This very large value results from an efficient mixing of the ionic configurations into the mainly singlet diradical ground state which is feasible because the semiquinonate SOMOs are delocalized and, therefore, have moderate on-site Coulomb repulsion parameters. As pointed out in the discussion, this represents an interesting difference to the case of magnetically interacting transition metal ions which typically show much smaller magnetic exchange couplings.