An ongoing research project at the Transportation Informatics Group at Klagenfurt University is concerned with the development of the advanced driver assistance system MIDCO - a "Machinevision based context-aware Intelligent Driver Co-Pilot". MIDCO will be an intelligent co-driver who will support the human driver and give advice in potentially dangerous and difficult situations. This thesis is embedded within the project and is mainly concerned with the development of a comprehensive knowledge base for the planned context-aware driver support system. Most of the active research in driver assistance is still concerned with the gathering of environment information. Much less attention is given to the question what shall be done with the information once it is available. The presented work will focus on this by trying to give an answer to the three major questions:- What is the context for driving assistance on a tactical level and how can it be represented? - How can traffic rules be integrated in a driver assistance system? - How can imperfect sensed context-information be handled?An approach for knowledge representation and processing of driving context and traffic rules will be developed and validated within the thesis which is also able to cope with imperfect context information.