The surface of a material is its connection to the outside world, and it is through the surface that a material can interact physically and chemically with its environment. Many of the interesting properties of functional materials are controlled by their surface structure and chemistry, requiring a careful understanding of the surface to understand and apply them. In particular, metal oxides are becoming increasingly important in applications such as catalysis and novel electronics. In this dissertation, the surface reconstruction of two complex metal oxide surfaces are investigated using a combination of experimental and theoretical methods, in particular density functional theory calculations performed with the WIEN2k code. The first system studied is the magnetite (001) surface, which undergoes a (2 2)R45 surface reconstruction that is capable of stabilizing single metal adatoms, making it an interesting model catalytic support. The mechanism by which adatoms selectively adsorb on this surface was not previously understood. Using a combination of STM and LEED IV experiments interpreted with DFT calculations, we have proposed a new structural model of this surface, which is stabilized by an ordered array of subsurface cation vacancies (SCV). This SCV structure achieves an excellent agreement with experiment, on par with well-understood metal reconstructions. The second system involved the study of two reconstructions of the strontium titanate (110) surface. This surface undergoes a wide variety of reconstructions depending on preparation conditions. Two reconstructions were investigated using XANES, crystal field multiplet and DFT calculations, with emphasis on the unusual tetrahedral coordination of Ti on the 4 x 1 surface reconstruction.