This paper took Xuzhou city, which is ongoing function transformation, as the study region, divided land resources by interpreting remote sensing images into 5 types, i.e., water, forest, construction land, farmland and unused land, calculated the probability of transfer matrix in 2001, 2005 and 2008, and then predicted the changes of land use in the coming 9 years with weighted Markov model which could improve the accuracy of prediction. The results showed that from 2009 to 2017 the area of construction land will increase 7.363 km2, the areas of water and forest lands will rise slightly, but the area of farmland will continuously decrease 9.61 km2.