Cognitive Radio Networks (CRNs) arise as a promising solution to the scarcity of spectrum. By means of cooperation and smart decisions influenced by previous knowledge, CRNs are able to detect and profit from the best spectrum opportunities without interfering primary licensed users. However, besides the well-known attacks to wireless networks, new attacks threat this type of networks. In this paper we analyze these threats and propose a set of intrusion detection modules targeted to detect them. Provided method will allow a CRN to identify attack sources and types of attacks, and to properly react against them.