<p>With the goal of addressing the issues of energy limitation in routing protocols as well as the problem of selfish nodes choosing to maximize their own interests to save energy and refuse to take part in the data forwarding task in real application scenarios, we proposed a non-homogeneous wireless sensor network clustering routing protocol based on non-cooperative game. In order to encourage each node to report its actual payoff and cost and calculate the equilibrium probability of becoming a candidate cluster head, we first assumed that all nodes were rational, self-interested individuals. Secondly, we modeled the election process of candidate cluster heads as a non-cooperative game. Alternatively, the election of cluster heads relied on the competitive radius mechanism, which resulted in more cluster head generation in the vicinity of the base station, achieved non-uniform clustering, and circumvented the issue of “hot spots.” In addition, we presented the energy Gini coefficient to quantify the energy consumption balance of the network. The simulation results demonstrated that, in comparison to other routing algorithms, the number and distribution of cluster heads produced by the suggested algorithm were more reasonable, and the energy consumption was more evenly distributed, which prolonged the life cycle of the network and improved the network throughput and packet delivery ratio.</p>