Probabilistic engineering mechanics aims to take into account uncertainties of input parameters of models and evaluating their impact onto the model responses. For frequency dependent visco-elastic sandwich structures, used in many industrial applications such as for vibration damping, the computation of the stochastic response can lead to a prohibitive computational cost. In this paper, an efficient methodology is proposed to analyze the output variability by a stochastic method based on fast Monte Carlo Simulation using a specific formulation obtained with the modal stability assumption. Geometrical and mechanical properties of a three-layered sandwich beam with a visco-elastic core made of 3M ISD112 damping polymer are represented by random fields to model the spatial variability. The validation of the formulation is studied, then the variability of the frequency response is evaluated for moderate and high variability levels. The benefits of the approach are also discussed.