This paper describes an efficient 3D modeling method from 3D range data-sets that is utilizing range data segmentation. Our algorithm starts with a set of unregistered 3D range scans of a large scale scene. The scans are being pre-processed for noise removal and hole filling. The next step is range segmentation and the extraction of planar and linear features. These features are utilized for the automatic registration of the range scans into a common frame of reference [Automated feature-based range registration of urban scenes of large scale]. A volumetric-based algorithm is used for the construction of a coherent 3D mesh that encloses all range scans. Finally, the original segmented scans are used in order to simplify the constructed mesh. The mesh can now be represented as a set of planar regions at areas of low complexity and as a set of dense mesh triangular elements at areas of high complexity. This is achieved by computing the overlaps of the original segmented planar areas on the generated 3D mesh. The example of the construction of the 3D model of a building in the NYC area is presented.