Track geometry quality is an important aspect in railway engineering as it reflects the actual condition of a track giving account of track geometry deviations. Monitoring and prediction of a relevant geometry quality parameter over time provides opportunity for effective maintenance with advantage of extending the life of the asset, reducing maintenance cost and minimizing possession time requirements. Two important aspects of good maintenance practice relating to track geometry quality are quality assessment of every measurement run for special and common cause of variations and also understanding the progression of the deterioration process. This gives engineering insight into temporal failure phenomena including the behaviour of track structure over time that can facilitate condition forecasting and consequent maintenance planning. This paper presents an approach for assessing track geometry data and also compares three track quality prediction models- linear, exponential and suggested GM(1,1) models. A series of inspection data from a selected line section of Trafikverket (Swedish transport administration) is used in the study. The contribution of this paper is the improvement of prediction accuracy of track geometry model, which is an essential consideration in failure prediction technique.