Résumé : We consider network clustering as the way to improve the performance of locating data in unstructured P2P systems. Connectivity-based Distributed node Clustering (CDC), and SCM-based Distributed Clustering (SDC) are two major protocols that allow partitioning a network topology into clusters, based on node connectivity. These protocols focus on the accuracy of the clustering scheme, i.e. using the Scale Coverage Measure (SCM), and its maintenance against node dynamicity. However, they do not propose search techniques that may take advantage of their clustering information. Thus, their proposals have not been evaluated according to the motivation behind. In this work, we propose a new, efficient Cluster-based Search Technique (CBST) for unstructured P2P systems. We use it to validate connectivity-based clustering schemes, according to the trade-off between cost of maintaining clusters, and benefit for query processing. Our experimental results show the efficiency of CBST implemented over the SDC protocol. By simply exploiting clustering features of the underlying network, a query can travel across a large number of nodes with a minimum number of messages. CBST eliminates a large portion of redundant messages, thus avoiding to overload the P2P network.