A prerequisite for studying catalytic processes at surfaces is a detailed knowledge of the surfaces' structural properties. Thereby the subject "structure" cannot be restricted to atomic relaxations alone, but also must consider substitutional ordering phenomena which often take place on a mesoscopic scale. Two examples whose modelling is fundamental for studying chemical processes at surfaces, are surface segregation and multi-site adsorption. It will be shown that today it is possible to study these properties using a first-principles based Hamiltonian constructed form the energetics of geometrically fully relaxed structures. This Hamiltonian can then be used for Monte-Carlo simulations in order to investigate the temperature-dependence of the surface properties. It will be shown that our new UNCLE code allows for a quantitative prediction of segregation, adsorption. and short-range order in excellent agreement with experimental data. Our focus will be on segregation at metal alloy surfaces and nanostructures stabilized by hydrogen adsorption.