In the last few years in Poland the railway infrastructure modernization program was lounged. It requires fast and precise technique to acquire data sets. Mobile laser scanning could be implemented, however automatic modeling methods from point cloud data sets are not suitable for geometrically complex railway infrastructure equipment such as traction poles. The main object of this study is the development of automatic traction poles extraction algorithm from laser scanning data. The flexibility of the method and independence from user-defined parameters were the main algorithm objectives. Because of the laser scanning data volume, simple calculations on point cloud subsets should be used to assure processing efficiency. In this study the combination of density and distance analysis was used. Proposed algorithm has been divided into two stages. In the first step regions of interest are selected by analysis of density difference for points located directly above the railway tracks. The influence of point density bin size on the number of correctly classified region was tested. In the second stage, each of the potential regions is analyzed individually. Iterative method of rejecting points based on distance criteria was used to extract traction poles points. In the study the point cloud from mobile laser scanner with density of 700 points/m2 was used. The test area covers 1.5 km railroad section between Miechow and Slomniki in Poland and contains 26 traction poles. All traction poles within study area were detected. It was proved that by appropriate combination of density and distance analysis, accurate traction poles extraction is possible even in complex regions with many surrounding objects.