Improvement of railway capability results in heavier axle loads and higher speed lines, which further induces railway subsidence. In order to ensure a good railway performance and reduce railway life cycle costs, railway subsidence should be measured regularly. The paper aims to assess railway performance by monitoring land subsidence along the railway, predicting railway subsidence in the future based on historical subsidence records. Persistent Scatterer Interferometric Synthetic Aperture Radar (PS-InSAR) is adopted in this research for monitoring land subsidence along the railway while Autoregression Moving Average (ARMA), artificial neural network and grey model are applied for subsidence prediction.