With the explosive development of wireless technologies, the demand of electromagnetic spectrum has been growing dramatically. Therefore, looking for more available spectrum, regulators have already begun to study secondary assignments in licensed bands. In this paper we present a probabilistic model based on a stochastic geometry approach to analyze cognitive radio networks. We focus on those scenarios where more than one band is available, a natural situation in this kind of networks. Quiet surprisingly, and to the best of our knowledge, such scenario has not been deeply explored yet in the literature. In particular we focus our study in the two main performance metrics: medium access probability and coverage probability. We evaluate our proposal through simulations and we present the analytical results of a particular case.