In current bridge maintenance practice, condition grades are assigned to individual bridges, based on regularly performed inspections. One of the main limitations to this approach is the subjective nature of grade assignment. To overcome this drawback, major bridge authorities are developing new methods for condition assessment based on collecting and evaluating sensor data. A major challenge in this context is to correctly model the impact of local deteriorations on the entire bridge's state. In this research, a system model-based approach has been developed to accurately model the correlations between the deterioration mechanisms and the measurement values indicating the progress of the deterioration. In addition, the system model describes the impact of the condition of individual bridge components on the condition of the overall bridge system. To this end, the bridge is hierarchically decomposed into modules, components and subcomponents, taking the structural system and mutual dependencies into account. The system model consists of three levels: The lowest level provides elements for modelling the input parameters provided by sensors or manual measurements. The mid-level models the deterioration mechanisms, taking the output of the parameter level into account. The topmost level models the structure of the bridge in a hierarchical manner, starting at the element parts up to the complete bridge system. The bridge´s condition is determined by state propagation mechanisms on the basis of logical elements connecting the aforementioned elements. In the end, the system model can be used to simulate the propagation of conditions assignments from the leaves (the sensors) to the top (the entire bridge). The developed system model approach is based on the application of the Systems Modelling Language (SysML). The paper will discuss in detail the advantages and limitations of the developed method and present a number of ostensive examples.