The article considers the problem of spatial objects recognition in geographic information systems. The solutions for this problem are given, a classification of the methods and algorithms already available is executed and the ways of development in search of the optimum decision on the matter are offered. Firstly, it is the improvement in the quality of recognition. Secondly, it is the creation of general rules of sufficient accuracy rating to prevent erratic classification of objects of a certain type. Thirdly, it is the establishment of rules defining correlations of spatial objects of different classes.Models and algorithms of spatially distributed objects recognition can be classified based on the following directions: with using external and internal topological relations, with application of neural network approach and mixed methods. Not only the issue of object recognition importance, but also the importance of accounting of their relative positioning and determination of spatial characteristics in relation to each other is considered.