The surface of the earth is changing rapidly, at local, regional, national, and global scales. Land use changes, especially those arising from human activities is among the most important global changes that impact the environment more than any other world is changing. The remote sensing technology as a valuable approach to monitoring, identifying and zoning of natural resources, especially in the process of preparing the land use maps in different parts of the world are documented scientifically. The maps are the basic requirements for environmental management and monitoring. Prediction and evaluation of potential land use patterns through modeling can help the environmental planners and natural resource managers to more conscious decisions on prospective planning. Therefore, in this study, mapping, change detection and land use change modeling using CA- Markov and LCM models in The Central District of Shiraz township has been. For this purpose, Landsat TM images for the years 1994 and 2009, ETM + in 2002 and OLI 2015 was used. At first, a set of initial preprocessing was applied to all images such as topographic, atmospheric and radiometric correction. Then, to make the best visual interpresion of satellite images, a set of indices including Optimal Index Factor (OIF) and vegetation index was used. Finally supervised classification method (maximum likelihood) was applied to the images and all images to obtain seven main types of LULC include bulid-up, agriculture, water area, barren land, forest, poor range and rich range with acceptable Kappa index and overall accuracy. Then, it was used from post classification comparison in LCM model to detect changes in this period. The most important changes that have occurred in this period 21 years belongs to Change 99/27 percent water area to barren land. The most important reason to convert water area, has been dry maharloo lake. That has caused social and ecological structure and physical stability of the region suffered serious damage. Also in this period, 15148 hectares has been added to an area of residential and bulid-up areas in study region .Cellular automata- Markov and artificial neural network (MLP) models was applied to predict LULC changes for 2015 (model calibration) and 2025 (future prediction). To implement neural network model was initially calibrated. For this purpose, by considering changes in the period 2002- 2009 and choosing 10 independent variable, potential landuse changes modeling for year 2015 was performed. Then, land use map for 2015 were predicted using Markov chain. after next, modeling land use map of 2015 compared with the reference map and Kappa index was obtained 0/78. Therefore, this model was used to predict changes for the year 1404. also To check the ability Cellular automata- Markov model, simulated map of the 2015 were compared with reference Map and Kappa index was obtained 0/81. Simulated maps of two models indicate that by the year 1404 will be significant growth in residential and bulid-up lands.