Abstract: To reduce the energy consumption for data query and prolong the lifetime of wireless sensor networks, a depth-first K-NN query algorithm based on filters is proposed(FKDF). In the FKDF algorithm, the K-NN query range is determined by setting filters for each node. As depth first traversal is carried out, the query information is formed by using adjacent list information of the query nodes. Then, a sink node delivers query information and waits for the query nodes sending back query results, thereby decreasing the average hops per query process needed consequently. Simulation results show that FKDF algorithm reduces the average hops per query process needed, compared with the FILA algorithm setting filters and the GPSR routing algorithm, and it can not only adapt to the dynamic network topology structure, but also realize the minimal query cost when the value of K changes constantly.