In our work we present multifractal formalism as a tool for description and extraction of information on very high spatial resolution satellite images. This approach is based on an assumption that single image (multifractal) consists of number of fractals, each with different dimension. Multifractals are used for description, modelling, analysis and processing of different complex shapes and signals. In particular multifractal decomposition can be used in the analysis of heterogeneous measures and structures typical for satellite images. It allows for detailed characterization (local and global) and description using functions. Listed advantages motivate our work on this topic. First part of our paper is a review of multifractal methods applied so far in remote sensing. Next we present our approach and results of analysis done on 159 subsets of images acquired by WorldView-2 satellite. Our test samples present different land cover types. Conducted analysis shows that generalised dimensions designated for individual fragments of images differs depending on the present land cover types. Also values of multifractality are connected to a land cover type. In general they allow for automatic assignment of land cover types to specific classes. Some deviations take place in case of discrimination between agricultural areas and forests - this will be a point for future investigation. The highest multifractality level can be observed for urban areas, the lowest for water that can be considered as a monofractal. Conducted analysis shows that multifractal formalism creates additional possibilities for the description and automatic classification of images.