Abstract: Based on the survey data of land use change in Zengcheng city, Guangzhou in 2005, firstly, we propose a method which uses three models including Logistic regression, artificial neural network and Autologistic regression to analysis the relationships between the land use types and their driving factors. Then, we combine the best goodness-of-fit of Autologistic regression model with CLUE-S model to simulate the land use pattern of Zengcheng city in 2009 and validate it using the actual land use data in 2009. Results show that the Kappa index of the simulation reaches 0.8637, an ideal result. Finally, we design two contextual modes: natural growth and optimization strategy, and simulate the regional land use pattern in 2025 based on the previous research, which can provide some decision-making references for the land use planning revision and future urban planning layout of this area.