Abstract: Sensor nodes in wireless sensor networks (WSNs) are usually powered by non-rechargeable batteries. In order to prolong the network lifetime of WSNs, it is necessary to save and balance the energy consumption of sensor nodes in WSNs. Therefore, energy-efficient routing of WSNs has become the focus of attention. In existing routing protocols, clustering routing is considered to be a very effective method. However, most of clustering routing algorithms randomly elect cluster heads and the position of cluster nodes within clusters is not carefully considered during clustering. The poor node position of aggregating nodes and the uneven distribution of energy consumption in WSNs seriously affect the network lifetime. To address the problem, an energy-efficient multi-hop routing protocol based on bifunctional nodes is proposed in this paper. Unlike other routing algorithms which have only a cluster head in a cluster, this protocol introduces a communication node as another functional node. Communication nodes are responsible for reorganizing clusters and relay inter-cluster routing, which reduces the overhead of cluster heads and balances the energy distribution of the networks. A non-uniform grid cluster division method is adopted to divide the network area into non-uniform grids which organize nodes in a grid into a cluster. A variety of factors, such as the node energy, the position and the level of grids, are synthetically considered in the protocol. The cost of communication among member nodes in clusters is reduced by reducing the complexity of the election and optimizing the position of functional nodes in the clusters. A multi-hop routing algorithm is proposed to construct class-spanning tree between clusters to reduce the complexity of electing functional nodes. Simulation results show that the proposed routing protocol improves the energy efficiency of WSNs. Compared with EECS, EEBCDA and multi hop EEBCDA, the network lifetime of the routing protocol is extended by at least 17.5% and the energy consumption of nodes is more balanced.