Abstract Global properties of Multi-Degrees-of-Freedom (M-DoF) systems, in particular phase space organization, are largely unexplored due to the computational challenge requested to build basins of attraction. To overcome this problem, various techniques have been developed, some trying to improve algorithms and to exploit high speed computing, others giving up to possibility of having the exact phase space organization and trying to extract major information on a probability base. Following the last approach, this work exploits the method of “basin stability” (Menck et al., 2013) in order to drastically reduce the numerical effort. The probability of reaching the attractors is evaluated using a reasonable number of trials with random initial conditions. Then we investigate how this probability depends on particular generalized coordinate or a pair of coordinates. The method allows to obtain information about the basins compactness and reveals particular features of the phase space topology. We focus the study on a 2-DoF multistable paradigmatic system represented by a parametric pendulum on a moving support and model of a Church Bell. The trustworthiness of the proposed approach is enhanced through the comparison with the classical computation of basins of attraction performed in the full range of initial conditions. The proposed approach can be effectively utilized to investigate the phase space in multidimensional nonlinear dynamical systems by providing additional insights over traditional methods.