Important structural and mechanistic details concerning the non-heme, low-spin Fe(III) center in nitrile hydratase (NHase) remain poorly understood. We now report projection unrestricted Hartree-Fock (PUHF) calculations on the spin preferences of a series of inorganic complexes in which Fe(III) is coordinated by a mixed set of N/S ligands. Given that many of these compounds have been prepared as models of the NHase metal center, this study has allowed us to evaluate this computational approach as a tool for future calculations on the electronic structure of the NHase Fe(III) center itself. When used in combination with the INDO/S semiempirical model, the PUHF method correctly predicts the experimentally observed spin state for 12 of the 13 Fe(III)-containing complexes studied here. The one compound for which there is disagreement between our theoretical calculations and experimental observation exhibits temperature-dependent spin behavior. In this case, the failure of the PUHF-INDO/S approach may be associated with differences between the structure of the Fe(III) complex present under the conditions used to measure the spin preference and that observed by X-ray crystallography. A preliminary analysis of the role of the N/S ligands and coordination geometry in defining the Fe(III) spin preferences in these complexes has also been undertaken by computing the electronic properties of the lowest energy Fe(III) spin states. While any detailed interpretation of our results is constrained both by the limited set of well-characterized Fe(III) complexes used in this study and by the complicated dependence of Fe(III) spin preference upon metal-ligand interactions and coordination geometry, these PUHF-INDO/S calculations support the hypothesis that the deprotonated amide nitrogens coordinating the metal stabilize the low-spin Fe(III) ground state seen in NHase. Strong evidence that the sulfur ligands exclusively define the Fe(III) spin state preference by forming metal-ligand bonds with significant covalent character is not provided by these computational studies. This might, however, reflect limitations in modeling these systems at the INDO/S level of theory.