This study proposes a fast algorithm for nose tip localization, which is robust to pose variations. Based on the coordinates of the local reference frame (LRF), the plane-distance energy of each vertex is calculated and a novel iteration algorithm for selecting candidate points is designed. For each vertex with centralized candidate points, the divergence on the three-dimensional (3D) vector field is computed. The nose tip denotes the point with the maximum divergence value. The efficiency of the algorithm is verified by applying it to the FRGC v2.0 and Bosphorus face libraries. The average runtime of nose tip location is only 0.62 s on the Bosphorus library, whereas the location accuracy is 95.6% on the FRGC v2.0 library. Finally, compared with other state-of-the-art algorithms, the proposed algorithm ranks the first both in speed and accuracy. The results show that the proposed algorithm can meet the requirements of real-time processing, has relatively high accuracy, and is robust to the pose variations in human faces.