Understanding track geometry deterioration decisively influences the planning and optimisation of track maintenance and renewal works and consequently the substantial related costs (savings) each year of every railway. To understand this deterioration it must be accurately modelled, paving the way towards its forecasting into the future. The main aim of the present study was to model railway track geometry deterioration using multivariate statistical analysis of the variables involved and to predict the future behaviour of the track geometry deterioration. For this purpose, a track section of about 180 km in length was selected as the base for the model and divided into analytical segments of as uniform characteristics as possible using a special segmentation algorithm. The lengths of the individual analytical segments were not identical, but were also kept as close to uniform as possible. For each analytical segment, the following general information was collected: track structure, traffic characteristics, track layout, environmental factors, track geometry measurements records and maintenance and renewal history data. Consequently, multivariate statistical analysis was performed on the main track geometry parameters: twist, level, alignment, gauge and cant. The coefficients of the independent variables involved in track geometry were found for each parameter in order to predict future behaviour for the purposes of efficient maintenance and renewal management.