In this paper we present a self-organizing, singlehop clustering scheme, which is based on partitioning sensor networks into several disjoint cliques. Clustering sensor nodes into small groups is an effective method to achieve scalability, fault tolerance, load balancing, routing etc. Here, we develop and analyze maximal clique based cluster-first technique where each node obtains a list of its neighbours' connectivity as well as their degree of connection at first. Then, the node with highest degree of connection initiates clique formation process and makes the cluster. Among all the members of the cluster, the node with maximum energy is selected as cluster head (CH). The proposed technique has a number of advantages. For example, it requires only the knowledge of one-hop neighbours to form clusters. Furthermore, the clustering algorithm is robust for topological change caused by node failure, node mobility, CH change and even for node insertion or removal. Simulation results show that our proposed clustering scheme gives better performance in terms of cluster size, variance of cluster size, and number of single node clusters than the existing clustering algorithms such as Secure Distributed Clustering, LEACH and LCA.