We consider a model of individual preferences in which the preference of each individual is characterized by a stochastic vector (stochastic model of preferences). Using a sociological survey or analysis of user actions in an online social network, it is possible to obtain the probability distribution of individual preferences. The problem of finding the median preference—a vector that minimizes the expected distance to the preferences of individuals—is posed and solved. It is shown that knowing partial (marginal) distributions is sufficient to find the median preference. Illustrative examples of finding median preferences for the three-dimensional case are provided.