In many regions both urban expansion and rural development take place simultaneously, and for the purpose of understanding the dynamic process of land use/cover change (LUCC) in such large areas, this study develops a multi—agent based land use model. Taking the Poyang Lake area of China as a typical case, this study applies the mechanism of diffusion—limited aggregation to simulate the behavior of urban agents, while rural land use is illustrated with a bottom—up based model consisting of agent and environment layers. In the agent layer, each household agent makes its own decisions on land use, and at each time interval a government agent takes control of land use by implementing policies. According to incomes and the rate of migrant workers, household agents are divided into six categories, among which different decision rules are followed. For complex LUCC in the Poyang Lake area of China from 1985 to 2005, the artificial society model developed in this study yields results highly consistent with observations. Importantly, it is shown that governmental policies can impose significant effects on the decisions of individual household agents on land use and the multi—agent—based land use model developed in this study provides a robust means for assessing the effectiveness of governmental policies.