Abstract: Dynamic multi-scale representation of the road network is required for navigation applications. In this paper, after depicting transportation features in thematic, spatial and temporal domains, a Feature-based Dynamic Multi-Scale Navigable data model (FDMSN) is devised with spatial, temporal and scale pictograms enabled UML class diagrams. In the model, firstly, topology and geometry of the road network are represented separately, where carriageways, which are constructed by linear referencing to underlying roads, are used as basics of the topologic network, and roads are used as basics of the geometric network. Secondly, these two networks are organized into hierarchical levels in terms of different criteria, that is, road class for the topologic network and display class for the geometric network. Thirdly, some different kinds of dynamic information, including traffic status data, traffic event data, statistical travel time data, etc., are integrated with the rock-bottom road network by link-node linear referencing systems and time-dependent thematic attributes. The experiment shows that the proposed model is effective enough to navigation applications by underpinning the corresponding data structure design and database implementation.