Track geometry tends to deteriorate over time under the influence of traffic load. This means that the track tamping needs to be carried out when the track geometry exceeds defined limits. Better understanding of this process is essential basis for efficient maintenance and renewal ensured by periodical measurements of the railway network, subsequent data evaluation and life-cycle cost analysis. Evaluation of the quality of track geometry is based on certain variables calculated for short sections of the track. The aim of the paper is to present a way to accurately rank quality of the track geometry using mathematical statistics and furthermore to be able to predict the progress of the track quality via regression analysis. These methods could be also utilized to help understand processes occurring in the railway substructure and superstructure.