Cognitive radio networks represent a new class of wireless networks where the channels are not permanently available, depending on the traffic activity of licensed users who have the priority to use these channels. We design a novel routing procedure for multihop cognitive radio networks composed of adequate metrics and a strategy to combine these metrics. The objective is then to increase channel availability when the routes are established. Two global metrics are defined. The stability metric evaluates the utilization efficiency of channels by capturing their sporadic availability to cognitive users. The predicted power metric estimates the spectrum capabilities for the on-going transmission without interrupting licensed users. We use fuzzy logic theory to compute and combine these metrics in order to make suitable routing decisions. Our procedure consists of two phases. In the first phase we compute the route to the destination, and in the second we examine the ability of this route to satisfy the required type of connection at the source. Numerical analysis and simulation results show that our procedure is able to find the route that goes through the nodes with better channel conditions. Fuzzy logic seems then to be an appropriate technique to decide the routes to establish in multihop cognitive radio networks.