Due to the significant costs and time consumed for track visual inspections, most railway industries rely only on geometry data obtained from automated inspections for the assessments of railway track quality conditions. This is the main limitation of the current practices, which may lead to inappropriate determinations of maintenance and repair schedules. This research attempts to rectify this deficiency by developing a methodology for the establishment of correlations between the track structural conditions and the data obtained from automated inspections. The aim is to provide the possibility of having a rational understanding of the structural defects of track (the causes of track irregularities) without conducting visual inspections. Neural network technique is implemented for this purpose. A vast amount of field data obtained from comprehensive visual and automated inspections of different railways are utilized to develop the neural network models. The results obtained in this research reveals that the neural network technique has a very good capability in establishing correlations between track geometrical defects and track structural problems. The application of the developed models in a number of railway tracks indicates that the proposed methodology is an effective approach in the prediction of track structural defects.