The point cloud data filtering is always an important problem in the research of airborne LiDAR data. A LiDAR point cloud data filtering algorithm based on region prediction is proposed. The method creates a regular grid with point cloud data and removes outliers, divides the experimental area into different blocks and uses sub-blocks′ elevation standard deviation to predict the terrain slope parameters, finally determines the ground points. The proposed algorithm has an advantage of obtaining threshold adaptively by the conditions of topographic relief of the region. The international society for photogrammetry and remote sensing (ISPRS) reference dataset is used to test the method. The experimental results show that the proposed method can effectively remove non-ground points, keep the ground points and is effective at minimizing total error rates.