Determination of snow cover depth in mountainous terrain is of major importance for avalanche monitoring systems. Besides it is needed as an input information for environmental analysis, especially in hydrology. The aim of researches addressed in this paper was to evaluate the feasibility of using terrestrial photogrammetric images of mountain slopes for point cloud generation for snow cover mapping. The test area was located in the Pięć Stawów Valley in High Tatra in Poland. The image acquisition was carried out for slopes between Gładka Pass and Walentkowy Wierch. The first set of images was acquired during the winter season, when the deep snow cover reaches its highest annual values. Subsequently the second set of images was taken in summer, after the snow cover melted. The terrestrial image network was formed from all the images. The bundle adjustment was calculated and the winter and summer point clouds were generated using the dense matching algorithm. The mesh was built using the adjusted summer images. Created mesh was treated as a reference surface for measuring height of winter points. Calculated heights were used as a measures of snow depth. For some parts of test area the automatic generation of point clouds failed due to low-contrast snow texture. In the rest of the test area the calculated snow depth is highest for the concave terrain formations. The results show that the terrestrial photogrammetry may by an attractive approach for acquiring the information about the snow depth distribution at small areas comprising slopes of mountains and valleys.