Abstract: Abstract: Based on the comparison of the evolution of land use structure between Beijing and Tokyo, we try to explore the goals of Beijing in land use in future and the quantitative goals in 2020 and 2030, and then optimize land use structure on amount and simulate spatial distribution, which can better develop the macroeconomic regulation and control function. Based on land use data in Beijing during 1992-2012 and in Tokyo during 1982-2011, we try to analyze the land use characteristics and evolvement rules in Tokyo, and then learn from it to develop the land use in Beijing, such as the direction of land use. Combining the land use and social economic development, we seek the detailed goals of land use in 2020 and 2030, and then establish the multiple-object optimization model that combines the economic, social and ecological goals. The Matlab2009 software is used to solve the relative optimal solution for the land use structure in 2020 and 2030. After that, we focus on yielding the optimized land use transition area matrix and the land use suitability map. First, we use the land use in Beijing in 1992 and 2012 to generate land use transition area matrices with an interval of 8 and 18 years, and then optimize the 2 matrices according to the principles of land use development in future, which pays more attention to reducing the frequency of land use conversion between land use patterns, and controlling the transition from farmland to construction land, and from ecological land to farmland. Second, we choose the natural, social and distance factors to yield the land use suitability map. The natural factor includes altitude, aspect and slope, the social factor contains gross domestic product (GDP) and population, and the distance factor mainly contains the distance to highway, railway, river, road, city, and original land use pattern itself. According to the optimized land use amount, the corresponding optimized land use transition area matrix and the land use suitability map, we employ the CA-Markov model to simulate the space distribution of land use. The research showed the optimized land use structure, and the proportions of farmland, ecological land, construction land, traffic land, water body and other land type are 21.00%, 50.50%, 19.00, 3.00%, 4.50% and 2.00% respectively in 2020 and 20.00%, 50.00%, 20.00%, 3.50%, 4.50% and 2.00% respectively in 2030. Farmland will decline and construction land and traffic land will increase in a reasonable degree, while ecological land, water body and other land type maintain as before. The optimized land use structure is in line with the goals of urban development, intensive utilization of land and ecological protection. In the validation process of the CA-Markov model, we not only verify the land use in amount but also in space distribution based on the Kappa coefficient. The absolute value of relative error on construction land is less than 0.3%, and the Kappa coefficients on amount accuracy validation are all above 0.6. Moreover, the spatial validation on Kappa statistics and a map comparison are also eligible. The model validation indicates that the land use optimization is sufficiently useful to planners and policy makers. From the optimization of land use in 2020 and 2030, the land use structure is gradually becoming stable, which means the changes in amount and space distribution are gradually becoming small, and also, the distribution of optimized land use is more intensive and ecological.