Contemporary Airborne Light Detection and Ranging (LiDAR) systems are capable to rapidly gather the data from large geographical areas with high precision and great density. As a result, obtained datasets can contain several tens of millions of points, making LiDAR data compression an important issue. In this paper, three domain-specific compression algorithms are compared against a general-purpose algorithm. Selected testing LiDAR datasets are derived from the practice to challenge common data compression issues. In this way, influences of the terrain type, point density, and number of contained points on the compression efficiency are studied.