Building vehicular networks in roads and highways is a challenging research topic with a large number of applications ranging from traffic jams and car collisions prevention to efficient route planning. The analysis of the distance between vehicles in roads is a key factor in, e.g., designing vehicular networks protocols or planning a supporting infrastructure to improve vehicular connectivity. This work proposes a Gaussian-exponential mixture model to characterize the time distance between vehicles in a highway lane, based on measurements collected at different locations in several highways of the city of Madrid, in Spain. The model arises from the observed behavior that some vehicles travel very close together, like in a burst mode, showing Gaussian inter-arrival times, while other vehicles are somehow isolated, showing exponentially distributed inter-arrival times. The experiments show that such a Gaussian-exponential mixture model accurately characterizes inter-vehicle times observed from real traces.