Quantum chemical calculations are employed to elucidate the origin of a puzzling diamagnetism for a hexacyanomolybdate(IV) anion, [Mo(CN)₆]²⁻, which was previously reported by Szklarzewicz et al. [Inorg. Chem., 2007, 46, 9531–9533]. The diamagnetism is surprising because for the octahedral (d)² complex one would rather expect a (paramagnetic) triplet ground state, clearly favored over a (diamagnetic) singlet state by an exchange interaction between two d electrons in the t_2g orbitals. Nevertheless, the present calculations reveal that the minimum energy structure of isolated [Mo(CN)₆]²⁻ is not an octahedron, but a trigonal prism; the latter geometry allows maximization of a σ-donation from the cyanides to the electron-deficient Mo(IV) center. Unlike for the octahedron, for the trigonal prism structure the singlet and triplet spin states are close in energy to within a few kcal mol⁻¹. Although the actual relative energy of the two spin states turns out to be method-dependent, the complete active space calculations (CASPT2; with the appropriate choice of the IPEA shift parameter) can reproduce the singlet ground state, in agreement with the experimentally observed diamagnetism. Moreover, magnetic measurements reveal a slight increase of the magnetic susceptibility with the increase of temperature from 100 to 300 K, suggesting an admixture of a thermally induced paramagnetism (possibly due to Boltzmann population of the low-energy triplet state) on top of the dominant diamagnetism. Our prediction that the geometry of [Mo(CN)₆]²⁻ should significantly deviate from the ideal octahedron, not only in the gas phase, but also in a periodic DFT model of the crystalline phase, as well as the experimentally confirmed diamagnetic properties, does not agree with the previously reported ideal octahedral structure. We suggest that this crystal structure might have been determined incorrectly (e.g., due to overlooked merohedral twinning or superstructure properties) and it should be re-investigated.