Abstract: The researches on 3DCM are in full swing over the world.By far, 3DCM can only express regular entities, such as common buildings, while it is difficult for them to express 3D complex landscape entities, such as cloverleaf junctions.The 3DCM system presented only posses functions such as browse and attribute query or can also fulfil simple spatial queries, at the same time, they are feeble to express topology relationships.The main reason is that the majority of the researchers consider 3DCM as one visualiation system, or even one CAD system, and the related main works focus on the mass data capturing, including DEM, building height and building texture.In addition, the seldom work has been done on data modeling, many 3DCM system are established on CSG model or CAD model.The objects in 3D city environment have some specialties, their expression must impose the third dimension information and reality. In order to express 3D entities in the city environment, we firstly analyze the features of 3D entities.Because the difficulty of the expression is how to express face objects and body objects.In order to simplify the expression, we assume that body object is composed of face objects.Regular face object can be expressed by one plane area or by one regular curved surface;therefore, we define polygon and function constructive surface (FCS) to express it.DSM can express irregular face object, and TIN is the best type to express DSM, so we define TIN patches to express irregular face object.On the other hand, many 3D entities, such as buildings, have regular shapes.For their regularity, their shapes can be expressed by a set of regular geometry primitives, such as, cuboid and prism, etc.As a result, we define polygon, function constructive surface, constructive solid geometry and TIN patches as the main multiple features to express face objects, body objects and DEM.Thus, we establish one 3D data model for 3DCM. The expression of spatial relationships between 3D entities is very important, so we must settle the expression of the spatial relationships to some degree.As almost all 3D-entities are located on the DEM that expressed by TIN, the spatial relationships between buildings and DEM can be established by TIN interposition, and some relationships between buildings can also be established through TIN.Spatial query is another very improtant question. R tree is verified to be one efficient data structure used to manage overlapped objects.It suits to the situation of the city environment, so we set up R tree index for 3D entities, in order to manage them efficiently. In order to verify the 3D data model, we explored one experiment system using OpenGL based on Windows 98 platform.Using this experiment system, we can construct a set of 3D objects including regular or irregular objects, even tree model.In addition, our experiment system can also process spatial query and measurement.The experiment proved that the 3D data model has some advantages, such as solving the expression of 3D complex entities, and also solving the spatial query and measurement operation.