Предложен новый метод передачи скрытых сообщений в социальных сетях на примере сети “Вконтакте”, позволяющий через структуру графа друзей пользователя внедрять секретные сообщения. Получены количественные оценки объема внедряемого сообщения в графы различного размера. Показана необходимость добавления избыточности во внедряемое сообщение. Представленный метод позволяет использовать другие графоподобные структуры социальной сети для внедрения скрытых сообщений. Purpose. This article addresses the construction of a new method for transmission of hidden messages in social networks. Methodology. The research employs methods of information theory, probability theory and mathematical statistics The Shannon entropy is used as the statistics for the analysis of an embedded message. Findings. The author proposed using the graphical structures of social networks as a container for the secret message transmission for the first time. As an example, the popular Vkontakte network is considered. The main idea of the method involves using the structure of the user’s friends graph to embed a secret message. Based on the available vertices (friends’ accounts), a complete graph is constructed, and its edges are enumerated. Each edge of the graph corresponds to one bit of the message being embedded: the bit is “1”, if the edge is present in the graph (one account in friends of the other), the bit is “0” if the edge is missing. To transfer the graph from one person to another, a key vertex is used. The specified vertex is connected by an edge with each connected component of the graph, which allows the graph to be transmitted using a single node and take into account all the vertices (including isolated ones). When retrieving a message, the key vertex and the edges connected to it are not considered. Conclusions. During the experimental research, it was shown that messages extracted from an empty container differ from the encrypted message by the probability distribution of bits. The necessity of adding redundancy to transmitted secret messages is shown with the purpose of “leveling” the statistical properties of an empty and filled container. The results of the experiment have showed that this method of steganography allows embedding a large amount of information into various social network structures represented in the form of a graph. It was noted in the paper that potentially “narrow” place of the algorithm is registration of new accounts. The restrictions imposed by the administration of some social networks for security purposes do not always allow automatic registration of new accounts, which makes the process of message embedding more difficult.