Summary: A two-stage discrete model for the location of facilities is considered. At the first stage, a set of facilities to be opened is selected. At the second stage, additional facilities may be opened due to the realization of random demand for products. Customers preferences are taken into account in choosing the facility in which they will be served. The quantile of losses (income with the opposite sign) is used as a criterion function of the model. Several optimization problems are stated. In the first problem, a set of facilities to be opened is selected for a given value of the reliability level. In the second problem, along with the set of facilities to be opened, the reliability level of the quantile criterion is selected. At the same time, restrictions on the level of reliability and the value of the quantile criterion are introduced. Two approaches to setting these constraints are proposed. To solve the problems stated, the method of sample approximations is used. A theorem on sufficient conditions for the convergence of the proposed method is proved. We formulate mathematical programming problems, the solutions of which under certain conditions are solutions to the obtained approximating problems. Numerical results are presented.