Since the performance of cooperative spectrum sensing in cognitive radio (CR) will decrease in the fading channel, a clustering cooperative spectrum sensing algorithm based on the cooperation in cluster and multi-clusters combination is proposed in this paper. In the proposed algorithm, all the CR users are divided into several clusters, and the CR user in each cluster with the shortest distance from the fusion centre is regarded as the cluster head. The cluster head collects the local sensing information from the other CR users in the cluster, obtains the cooperative sensing result of the cluster, and then sends the result to the fusion centre which will combine the results from all the clusters by "OR rule" in order to obtain the final decision on the presence of the primary user. The simulation indicates that when the channels are perfect, the performances of the proposed algorithm and traditional algorithm without clustering are nearly the same, however, when the channels are in deep fading, the performance of the proposed algorithm is improved remarkably, and it can be increased by the decreasing of the number of clusters. 