 We developed an effective and efficient method for point-wise semantic classification of 3D point clouds. Our approach handles unstructured and inhomogeneous point clouds, such as those derived from static terrestrial lidar or photogrammetry, and is computationally efficient, allowing us to process point clouds with millions of points in just minutes. The key to this efficiency lies in carefully defining neighborhoods around each point, which allows us to create a feature set that is both expressive and fast to compute. We evaluated our method on benchmark data from a mobile mapping platform and on several large terrestrial laser scans with widely varying point densities. Our results show that our feature set outperformed the state of the art in terms of per-point classification accuracy, while also being significantly faster than other methods. This article was authored by T. Hackel, J. D. Wegner and K. Schindler.