Landslide is one of the most costly catastrophic events in terms of human lives and infrastructure damage, early warning monitoring for landslides becomes more and more important. Especially, the monitoring for landslides in man-made structures, such as bridges, dams and hydraulic engineering, plays a key role in the prevention and mitigation of risks related to natural and technological hazards. In most cases the occurrence of landslides is originated by the loss in equilibrium of the soil mass due to changes in one or more parameters, such as seismic noise, ground displacements, piezometric level and rainfall, which contribute to leaving the mass itself stable. Through continuous monitoring the correlative parameters, the dynamics activity of the landslide can be observed. In the past twenty years, many monitoring methods were proposed, but the dominant method is displacement amount testing, such as mechanics-electronics displacement measuring, topographic surveys and GPS surveys. These traditional monitoring techniques give information about displacements only in a few numbers of points. In addition, it takes much time and money for a large number of points to measure on a continuous basis. In fact, the landslide will occur when the balance between the hill’s weight and the countering resistance forces is tipped in favour of gravity. So, if the intra-stress distribution and changes of the landslide bodies can be monitored, the occurrence of landslides will be predicted accurately. With particular advantage, a distributed fiber optic stress sensor can be used to monitor the intra-stress distribution and changes in landslide bodies. According to the characteristics of the intra-stress distribution and changes detected by the distributed fiber optic stress sensor, combining with some applications of specific mathematical models, the movement the landslides can be predicted, thus the disaster can be avoided.