The presented work is devoted to the development of an approach to the age inference for users on the online social network VK.com. This approach is based on an analysis of the communities that the user is a member of. The proposed method is based on the hypothesis that users join some of those communities that contain data about class, graduation from school, data about the group at the university, etc. and those that have users’ peers as members. The proposed approach allows for the age inference. The obtained results can benefit research on online social networks, as well as the task of discovering a user vulnerability through social networks.