Optimizing service charges has become an important task for enterprises. Since these costs represent a major share in a companyís cost structure it is very important to be aware of the origin and cost drivers of this factor. Therefore the analysis of service charges is inevitable but often complex because of the heterogeneity of buildings and the lack of consistent data. The primary objective is to identify the main characteristics of buildings that influence operating costs. Hence this paper seeks to fill this gap by applying additive mixed regression models and a linear SUR-Model on a dataset containing detailed service charges for 1,400 German buildings located in 94 cities retrieved from 2000 to 2005. In a first step, service charges are estimated separately using semiparametric additive models with random city and firm effects. Based on these results, the models are then re-parameterized in order to achieve linear models. In a further step, we estimate a seemingly unrelated regression (SUR) model for these linear models. The results of this study can be used for ex ante as well as ex post analysis of service charges and are of great interest for owners, investors and operators of office buildings. The cost structure of any given building for example can be compared objectively to the results of this study and therefore reveal optimizing potential. The use of statistical methods makes the data comparable and transparent and therefore improves usual benchmarking approaches.