Wireless networks are starting to be populated by interconnected devices that reveal remarkable hardware and software differences. This fact raises a number of questions on the applicability of available results on dependability-related aspects of communication protocols, since they were obtained for wireless networks with homogeneous nodes. In this work, we study the impact of heterogeneous communication and computation capabilities of nodes on dependability aspects of diffusion protocols for wireless networks. We build a detailed stochastic model of the logic layers of the communication stack with the SAN formalism. The model takes into account relevant real-world aspects of wireless communication, such as transitional regions and capture effect, and heterogeneous node capabilities. Dependability-related metrics are evaluated with analytical solutions techniques for small networks, while simulation is employed in the case of large networks.