Ab initio quantum chemical modeling is used to predict differences in electronic structure between the (100) surface and bulk of pyrite. The density functional generalized gradient approximation (CASTEP) is used to obtain the density of states for the valence band of pyrite bulk and surface states. For this purpose the pyrite (100) surface is cleaved along all possible layers of the bulk, thus yielding a bulk terminated surface, a surface with the S-S dimer bond broken, and an irregularly cleaved fractured surface with exposed surface monomers, respectively. Furthermore, the hybrid functional B3LYP (CRYSTAL03) is used to predict S 2p and Fe 2p core line shifts for pyrite surface states. Experimental x-ray photoelectron spectroscopy (XPS) data for these core lines show the presence of two kinds of S surface states: surface S2- monomers at a S 2p(3/2) binding energy (BE) of 161.2 eV, and (S-S)(2-) surface dimer states at a S 2p(3/2) BE of 162.0 eV, compared to the S 2p(3/2) BE of bulk pyrite at 162.7 eV. Investigation of the ab initio valence band density of states for these different surfaces supports our conventional and synchrotron radiation XPS data, collected for the pristine pyrite surface, which suggests the presence of bivalent S monomers and monovalent S dimer surface states. The Fe 2p surface XPS displays several multiplets (implying high spin configuration) to higher binding energy than the bulk Fe 2p signal, which can be ascribed to surface state contributions. The quantum chemical simulation predicts an S 2p core level shift of 0.69 eV between the S bulk and S surface dimers, in good agreement with the 0.6 eV found in XPS measurements. Evidence for surface Fe2+ and Fe3+ surface states can be seen in the Fe projected valence band density of states, also confirming the interpretation of the photoemission spectra. From the B3LYP calculation, the surface contributions to the Fe 2p BE peak lead to a broadening of the signal, as observed by experiment.