Abstract: Since 1998 development of western China，the economic development of western China has been developing very rapidly，the western region land use taking place the tremendous and unprecedented changes，and profoundly changing the local ecosystem structure. Based on two phases TM images acquired in 2000 and 2010 respectively，making use of the technology of Remote Sensing and GIS，also integrating with CA-Markov model， a quantitative analysis was made of the quantity and spatial feature changes of land use in Guanzhong-Tianshui region during the past ten years and the future ten years，with a hope for providing the supports for the sustainable land use，land management decision-making，regional land and ecological environment construction，and a series of eco-environmental after-effects. In this study，the BP neural network classification algorithm was used to classify the two period’s remote sensing images and analyse the land use change transfer matrix. In the past ten years，LUCC was expressed as follows: the build-up land increased sharply and enlarged about 291 210 hm²，whereas the arable land and grassland decreased apparently，arable land decreased about 60 028 hm2 and grassland decreased about 374 770 hm². The build-up land has been increased in the way of invading the massive arable land，and the arable land reduction actually resulted in the reduction in per capita holding the average arable area，the grain question began to occur prominently; since this decade，about a quarter of the land has been changed in the study region， so human activities have developed strongly. And then the CA-Markov model was applied to carry on the simulation and prediction of 2020 land use，with the overall accuracy of simulating spatial location of 75.05%. In the coming ten years，LUCC will be listed as follows: Arable land and grassland will continue to be decreased and the reduction rate of cultivated land become greater in amount，build-up land area will still be in increasing rapidly and enlarged about 273 902 hm²，and most of them will be transformed from farmland and grassland，in the next ten years the actual changes in land use area will account for 11.43% of the region's total area. CA-Markov model with simulation time and spatial pattern change can reflect the changes in land use landscape patterns and their complexity and randomness with high precision and accuracy. More profound，intuitive study reveals the extent of environmental degradation. Land use and land cover change are a complex system，which is affected by many comprehensive factors，including both socio-economic and natural and environment ones.  Using the CA-Markov model as a basic tool to predicate land use/land cover may not well indicate the real facts，it still needs other methods to consummate the model in order to make it more approach to the real fact. Finally，the landscape changes of the study area were analysed by calculating landscape indices. Comprehensive analysis of the landscape index shows the trends in landscape fragmentation，complexity and uniform development and overall landscape pattern’s increasing heterogeneity; agricultural land is the most important type of land，because of its area reduction，the importance of the effects of agricultural land on overall landscape also decreases. The reduced agricultural land was changed into the other type’s lands，and some ratios among the land use types’ were reduced with increasing land use types’ diversity and uniformity. The reduction in landscape dominance index means decrease in landscape stability，thereby the withstand capabilities of landscape decrease.

Key words: land use, change matrix, CA-Markov, landscape ecology, Guanzhong-Tianshui region