Radio Frequency Identification Technology (RFID) is one of the key technologies of indoor positioning. The traditional LANDMARC location algorithms have poor positioning accuracy. To solve this problem, a novel location algorithm is proposed by combining the dynamic centroid iteration and error correction. It updates nearest neighbors in turn by employing the centroid of the neighboring area as next reference tag which takes the minimum location relationship as the criterion, and achieves the pre-positioning coordinate until the location relationship with the target tag is lower than the threshold. The correction factor is adopted to compensate error of the pre-positioning coordinate by relocating each of k-nearest neighbors. Simulation results show that the proposed algorithm performs better in terms of positioning accuracy than LANDMARC.