Multi-temporal satellite images (MTSI) are considered as the recent progress technology for remote sensing. This chapter referres to this context to benefit from the information extracted from MTSI to detect changes in vegetation. A wealth MTSI using large areas is now available for researchers. MTSI continue to play a significant role in understanding our planet. They have been considered in many applications such as precision agriculture, geology, environment and terrestrial ecology, urban growth and forest monitoring. Accurate and up-to-date land use data is necessary to understand and assess the changes. This issue is the subject of several research stemming from different points of view. In the literature review, change prediction methods proposed are often encountered problems within MTSI because they are completely influenced by seasonal and random variations. Two significant challenges stand out. First, methods must allow change detection in non-stationary MTSI. Secondly, methods must build a prediction system to predict future state of land use. In this chapter, an overview and comparative study of Hidden Markov models (HMM) used in land use change prediction is presented. In fact, HMM have been widely used in change prediction in MTSI under stationary hypothesis. Then a generic change prediction approach for MTSI, involving detection and characterization of change types (trend, seasonal and random) is provided. The experimental results on a series of LANDSAT images showed a good results for prediction. The experiment on MODIS dataset also confirms the improved performance of the propose system. More specifically, in section one, we motivate the use of remote sensing for change prediction, and give an overview about foundations for HMM in this context and its applications. Section two, describes the proposed approach for prediction changes with stationary/non stationary multi-temporal satellite images using hidden Markov model. Finally we will discuss the implementation results of the used technique and analyses the results within urban and vegetation context. We conclude with future works.