LIDAR provides precise information on the objects surveyed by sequentially measuring with high accuracy. The result is a georeferenced point cloud. Every point can be described by its coordinates, attributes, and its accuracy. However, the raw point cloud is not the final product. In order to retrieve information from the point cloud additional processing is applied like classification, filtering, and modelling. LIDAR systems with focal plane arrays acquire hundreds or even thousands of range readings with a single laser pulse. This provides intermediate point clouds with a tremendous point density. However, these point clouds are frequently extremely noisy. First of all, due to the high sensitivity of the detector down to the single-photon level, the point clouds show a lot of “points in the air”. Secondly, the ranging in itself is prone to a significantly higher level of range noise compared to linear waveform LIDAR. In order to still retrieve useful information these point clouds have to be pre-processed in order to reduce noise significantly. While “points in the air” can be disposed off by applying spatial density analysis, reducing range noise can only be tackled by some sort of spatial averaging. We discuss the impact of pre-processing of raw point cloud from focal-plane LIDAR with respect to changes applied to the information content and in comparison with point clouds delivered by state-of-the-art linear waveform-processing LIDAR systems.