Due to discontinuousness and incompleteness of point cloud data of potential unstable rock mass in high and steep slopes acquired from scanning equipment, there exist large errors in the extracted curvature feature of slopes and boundaries between unstable and stable rock mass are also blurred. To address those problems, a method for identifying unstable rock mass in high and steep slopes based on point cloud 3D reconstruction is proposed. Kriging interpolation is used in processing sparse or unevenly distributed point cloud data. More continuous and complete point cloud data can be obtained by increasing the density of data points in consideration of spatial correlation, so as to accurately reflect the real geometrical configuration of the slope. The normal vectors and curvatures of the point cloud data are estimated by adopting local surface fitting and k-neighborhood search algorithm, so as to extract the geometric features of slope and reveal the slope morphology and potential unstable areas. The inconsistency index is used to analyze the changes in the geometric features of point cloud data, and a threshold is set to screen edge points, thereby the region with unstable rock mass can be accurately delineated after the boundary between unstable and stable rock mass is clearly defined. The point cloud data within the boundary of the unstable rock mass are obtained by incremental Voronoi diagram, and then a 3D model of point cloud data is constructed for the unstable rock mass, and the risk level of the unstable rock mass is also analyzed with the Slide software. This is the whole identification process for unstable rock mass in high and steep mine slopes. The practical engineering application of this identification method shows that the slope curvature can be estimated with an error less than 0.3%, and the calculated F1 scores in identification of unstable rock mass in each group of slope all remain above 0.99, indicating that such method can bring high quality identification of unstable rock mass.