Abstract: We presented a model integrating GIS,cellular automata(CA) and genetic algorithm(GA) for urban land-use optimization(ULOM).Land use suitability,spatial harmony,amount proportion,future urban growth,slope,policy and urban planning,and so on,were all synthetically considered in this model,which was applied to optimize the urban land use pattern of Lanzhou with two different scenarios.The results show that the spatial replacements between residential and industrial land use in Chengguan District and Qilihe Distrist of Lanzhou should be picked up.Commercial land use in Zhangye Road,Baiyin Road and Yantan Road should continue to be strengthened.Land use growth for government and institution will centralize in Yingmentan of Anning District and Yantan of Chengguan District.There will be fast growth of residential land use in Matan and Datan of Qilihe District,and Yingmentan of Anning District.