Maintaining the Terrain Analysis System (TERAS) is the principal task of military geographic service. The system.s main component is Terrain Database. It is a huge set of raster, vector and attribute data. Military analysis of this input data should define: .cross.country mobility, . cover and concealment, line of sight, intelligence reconnaissance elements. This paper describes our experiences in the analysis of image data for military purposes. The analyzed dataset contains panchromatic and multispectral satellite scene: LANDSAT, SPOT, IRS, IKONOS and aerial photos. Different terrain and spectral resolution enables to distinguish from the imagery different quantitative and qualitative features. So, it is also an attempt to show the correlation between resolution and extraction of different kind of information. The basic military product is terrain evaluation map. It integrates information about military elements of geographic environment. Terrain database contains thematic layer data, for example: vegetation, hydrography, soils, relief etc. However, if we consider military aspects, this data could be grouped in three products: DEM, landcover map, soil map. Development of acquisition and processing techniques allows to create such maps from imagery. Is it possible to create terrain analysis subsystem only from digital imagery? This analysis tries to respond to this question. Such a subsystem would enable military terrain evaluation even if we have no others data. But first of all, we must gather the data and elaborate analysis and extraction algorithms. The main tasks are: optimal spectral composition analysis, .texture analysis and other methods of processing, continuous-to-thematic raster layer processing, advanced classifying procedures, conditional processing of aggregated data, proper ungrouping and composing output thematic layer. Imagery acquisition systems, especially satellite ones are constantly developed, and spectral and terrain resolution is increased. Therefore, we can obtain more and more information and we should search for new processing methods. The imagery data is especially valuable for military use of environment, because it helps to achieve dominance on the battlefield.