The paper presents the generalization methodology for data contained in digital terrain model (DTM) for the purpose of relief presentation on new topographic maps. An appropriate generalization process of digital terrain model is especially significant for geographic information systems (GIS). It is particularly important for reliable spatial analysis, therefore, to retain real position of characteristic points defining key landforms. It was decided that the proposed generalization process of digital terrain models was based on the hybrid structure of DTM generation from the most accurate height data in geodetic and cartographic resource. For this reason special application was developed which uses: an extraction of characteristic points and regularly distributed complementary points, as well as an extraction of structural lines to create a secondary DTM during generalization. The implementation of the whole process was carried out in Model Builder in ArcGIS 10. DTM created for a selected level of generalization was used then to generate contours for an appropriate scale, presenting the terrain on topographic maps. Two data source was used in presented research i.e. DTM from aerial laser scanning obtained in ISOK project and DTM created from photogrammetric data acquired in LPIS project. The proposed algorithm was tested on five areas representing different types of terrain: old glacial and young glacial landscape, low and high mountains and urban area. The results were compared for both data source. The statistical parameters were also calculated for the selected level of DTM generalization with regard to the raw data. For each test area DTM from LPIS and ISOK projects were positively generalized to hybrid structure which allowed for effective contours creation and acceptable relief representation on topographic maps in selected scale. Only few percent of initial data guaranteed appropriate cartographic accuracy in relief representation in result. As a conclusion authors indicate that full automation procedure of DTM generalization is not possible for whole Poland because of large morphometric diversity of our country. Presented research proved that achieving optimal results and satisfied cartographic accuracy can be obtained with usage of high resolution height data, exposed to the generalization algorithms, and supplemented by the current data from topographic data bases (TBD, BDOT).