3D models of the world offer a much richer environment (as compared with 2D models) in many different application domains (for example land development, emergency services, telecommunication, tourism or pedestrian navigation). There are benefits to using 3D models beyond just the visual. But there is no such common models exist so far to accomplish such tasks. Primary obstacle is this way is lack of 3D standards. In this regard, CityGML attempts to provide a rich description of 3D worlds sufficient to support semantic modelling and complex query. This research examines how the behaviours of pedestrians can be modelled in cities using CityGML to provide the underlying semantic model. CityGML supports the storage of information at different granularities. At the highest level of detail LOD4 shows the interior of a building, through to broad descriptions of terrain at LOD0. CityGML which is an extension of GML 3 application schema, its storage in the data base will be examined how a data model at different LODs can be stored internally in a database. The interaction between pedestrian and virtual city objects is to be examined and how an object is described by a pedestrian. This research highlights the benefit of using CityGML which is offered by this rich data model.