Nowadays, interactive driving simulations are well established to support the design of modern vehicle systems, like driver assistance systems (DAS). They enable to validate properties and functions of DAS in the early development phases. Driving simulations consist of several simulation models that work together at runtime. These partial models represent dedicated aspects of the vehicle, individual vehicle systems as well as the environment. In this context, environment models describe the environment the vehicle is driving through, comprising roads, buildings and vegetation. While models of single vehicle systems are currently systematically built with a high level of detail, building virtual environments is still a mostly manual and time consuming process. Particularly, building environment models that represent specific real world roads is laborious — geographically referenced data has to be gathered and integrated. Geographic information systems (GIS) and their underlying data — geo-data — have great potential to accelerate the generation of environment models. In order to achieve this potential, a systematic integration of geo-data is necessary. During this process, the specific requirements of interactive driving simulations have to be considered. This paper describes a method for the automated generation of virtual roadways based on geographic information systems. The method has been prototypically applied for a concrete road course and the results are briefly presented in this paper.