Introduction. As a rule, the quantitative assessment of the probabilistic characteristics of multicomponent risks involves the analysis of a large amount of statistical data, but quite often the available sample datasets are not representative or do not meet the requirements for the organization of systematic statistical accounting. Therefore, the scientific substantiation of the admissibility of using expert methods for quantitative assessments of the probabilistic parameters of rare events and phenomena is one of the urgent problems in the field of risk analysis and safety modeling of sociotechnical systems. The purpose of the study is to construct a numerical method for determining the probabilistic parameters of a stochastic model of multicomponent risks based on the procedure of paired comparisons by experts of the possibilities of various outcomes when a threat is realized and accepting a hypothesis about a functional relationship between the results of subjective assessment and objective measurement of probabilities. Methods. When constructing a numerical method, the procedure of pairwise comparisons is performed by experts for particular conditional probabilities of the occurrence of various consequences of the impact of hazardous factors on protected objects related to the given options for performing a set of protection functions. Determining a specific type of functional relationship between the results of a subjective assessment and an objective measurement of probabilities is possible on the basis of a retrospective forecast with the selection of a parameter that achieves the best correspondence between the simulated damage and real statistical data. Results and discussion. A numerical method is proposed for determining the probabilistic characteristics of multicomponent risks, based on the objectification of expert estimates of the probabilities of possible outcomes when exposed to hazardous factors on protected objects. The method of calculating subjective probabilities and their representation in a rank scale with further conversion into numerical values of the scale of ratios of the corresponding objective probabilities is substantiated. Findings. The application of the proposed numerical method for determining the probabilistic characteristics of multicomponent risks based on expert assessments is possible provided that a functional relationship is set between the results of a subjective assessment of the possibility of outcomes and an objective measurement of their probabilities in accordance with Stevens' psychophysical law. Keywords: expert assessments, multicomponent risks, psychophysical law, measurement theory.