With the development of information technologies, intelligent vehicle and driver support systems are started to take place more in our daily life. Such systems can warn the driver or even control movements of the vehicle by watching situation of roads and traffic signs. This paper introduces an approach which can be a part of such systems to detect and recognize traffic signs automatically. The video images taken from a vehicle in Turkish Highways are processed to find probable regions of traffic signs. After finding probable regions of signs in each video frame, the type of sign is determined by using features of traffic signs in the database. With the proposed approach, signs were recognized in various distances, scales, and viewing angles. Speed performance of the approach is sufficient to implement a real-time system in the future.