Metal oxides show a wide range of physicochemical properties, and play a major role in emerging and developing technologies. In all applications, surfaces are critically important: Devices are often driven by reactions and by interactions occurring at the materials’ surfaces. One befitted approach to explore and exploit the manifold properties of metal oxides, and to model their interaction with the environment, is to work with single- crystalline samples in ultra-high vacuum (UHV). This surface science approach provides tight control over sample composition, environment, and surface structures, thus meeting the needs of computational modeling, which, in turn, can offer interpretation and guidance to the experimental work. Bulk-like epitaxial thin films are a sound alternative to single crystals, especially when these are impure, too small, too expensive, or simply not available. Such ideal films can in principle be realized by various deposition techniques, among others pulsed laser deposition (PLD). However, numerous works from the literature testify to the challenges intrinsic to controlling the growth of metal oxide films by PLD: Films with rough morphologies, pronounced nonstoichiometry, and unexpected properties are produced more often than desired. This is largely because many growth parameters, e.g., the laser energy density, the oxygen background pressure, and the substrate temperature, have direct consequences on the film thickness, composition, crystallinity, morphology, and all the related properties. Moreover, non-idealities become more pronounced as the film thickness increases, making it hard to realize bulk-like samples of a few tens of nanometers. Multicomponent oxides are most severely affected, since their different cations suffer, e.g., from preferential ablation at the target, preferential scattering with the background gas, and preferential sticking at the substrate. The challenges in producing ideal, bulk-like oxide films are exacerbated by the difficulties in reproducing PLD parameters in different setups: Every PLD chamber is designed uniquely, and the relevant parameters can be measured in various ways. As a result, tools are needed to identify and tune the relevant growth parameters and achieve the desired film properties. This Thesis collects various studies on metal-oxide films, which have been produced and characterized with a combined PLD and surface science approach. In all cases, the aim was to produce single-crystalline films with fully characterized surfaces, suited to address the fundamentals of the surface reactions of interest. Surface science tools, most prominently scanning tunneling microscopy (STM), were used to explore the fundamentals behind the strongly parameter-dependent growth of In2O3(111)—a transparent conductive oxide used in various catalytic and gas-sensing applications—and of SrTiO3(110) and La1−xSrxMnO3(110), both perovskite oxides used as catalysts in solid-oxide fuel cells. In the case of In2O3, it was found that the oxygen stoichiometry of the deposited species, affected by the deposition conditions, strongly influences the surface diffusivity and, as a result, the film morphology. On the other hand, the studies on SrTiO3 and La1−xSrxMnO3, both characterized by composition-related surface reconstructions, have demonstrated that nonstoichiometries introduced under non-optimal growth conditions tend to accumulate at the films’ surfaces, forcing a change in the surface atomic structure. These changes are the potential cause for undesired morphology alterations, and may be a more general trait of perovskite oxides. It is shown how the newly developed relations between nonstoichiometry, surface morphology, and surface structure can be used to optimize film growth, and obtain stoichiometric, bulk-like, atomically flat films. A variety of surface science techniques has been used to thoroughly characterize the surface atomic details of the optimized films, with emphasis on La1−xSrxMnO3(110): Its surface reconstructions are presented here for the first time, and organized in a two- dimensional experimental surface phase diagram as a function of the oxygen chemical potential and the cation composition. The same tools have been used to characterize the surfaces of the more easily grown TiO2(001) anatase and Ti-doped Fe2O3(1-102), both interesting for their potential as catalysts for photoelectrochemical water splitting. In the case of Ti-doped Fe2O3(1-102) and La1−xSrxMnO3(110), the experimental surface characterization has been complemented by ab-initio theoretical calculations performed by collaborators. The importance of building model systems in the form of well-characterized single- crystalline surfaces to investigate given chemical reactions is exemplified by the studies on the oxygen incorporation on SrTiO3(110) and La1−xSrxMnO3(110) surfaces. Oxygen incorporation is a key process in solid-oxide fuel cells, but little is known on the exact mechanisms driving it. Using the well-defined model has enabled to shed light on the process, at the same time highlighting the critical role of surface atomic coordination and arrangement.