Traditional energy-balanced clustering algorithms in wireless sensor networks (WSNs) usually overlooked the different characters of energy consumption among clusters, which greatly influenced the efficiency of energy employing. To solve this problem, in this paper, a distributed power proportional clustering (DPPC) algorithm is proposed, in which the clusters are determined to be an optimal distribution by making the total power consumption of the cluster proportional to the total energy it stored. In the process of cluster formation, a distributed way is adopted. The sensors firstly compare their positions and organize themselves into the optimal clusters, then compete for the cluster heads according to their position and residual energy that are evaluated by the appropriateness function. The clustering topology formed by the DPPC makes the clusters have similar lifetimes with fully energy utilization, thus the energy efficiency can be greatly improved for the WSNs.