Intervehicle communication (IVC) is an emerging topic in research and application that is getting increasing attention from all major car manufacturers. In this paper, a novel method for scalable information dissemination in highly mobile ad hoc networks is proposed: segment-oriented data abstraction and dissemination (SODAD). With SODAD, information can be distributed in an information range multiple orders of magnitude larger than the transmission range of the air interface, even if only 1%-3% of all vehicles are equipped with an IVC system, e.g., during market introduction. By restricting the method to the dissemination of map/position-based data, scalability is achieved. In the second half of this paper, an example application for the SODAD method is presented: a self-organizing traffic-information system (SOTIS). In SOTIS, a car is equipped with a satellite navigation receiver, an IVC system, and a digital map. Each individual vehicle collects traffic information for its local area. Using the digital map, the traffic information is analyzed based on road segments. By distributing the information in the ad hoc intervehicle network using the SODAD method, a decentralized traffic information system is created. The performance of the proposed methods is evaluated using network simulation with vehicular mobility models. Simulation results for typical scenarios are presented. Furthermore, a prototype implementation based on commercially available standard hardware demonstrates the feasibility of the proposed approach.