In order to improve the energy efficiency of Wireless Sensor Network (WSN), a WSN routing algorithm based on Optimal Number of Clusters and Improved Gravitational Search (ONCIGS) was proposed. Firstly, the optimal number of clusters was calculated according to the idea of uneven clustering, and the improved AGglomerative NESting (AGNES) algorithm was adopted to realize the reasonable clustering of network. Secondly, reverse learning mechanism and elite strategy were introduced into the gravitational search algorithm, and the force was adjusted adaptively based on population density to improve the search precision and speed up the convergence. Then, the standard deviation of residual energy of cluster heads was taken as the objective function to search the energy-balanced inter-cluster data forwarding path. The experimental results show that, compared with the Low Energy Adaptive Clustering Hierarchy (LEACH) routing algorithm and Distributed Energy Balanced Unequal Clustering (DEBUC) routing algorithm, the network life cycle of the proposed ONCIGS is prolonged by 41.94% and 5.77% respectively under the network scale of 100 m×100 m, and it is prolonged by 76.60% and 7.82% respectively under the network scale of 200 m×200 m. The proposed ONCIGS can effectively prolong network lifetime and improve energy efficiency.