A new method to simulate urban land-use dynamics is proposed based on multi-agent systems (MAS). The model consists of a series of environmental layers and multi-agent layers, which can interact with each other. It attempts to explore the interactions between different players or agents,such as residents, property developers, and governments, and between these players and the environment. These interactions can give rise to urban macro-spatial patterns. This model is used to simulate the land-use dynamics of the Haizhu district of Guangzhou City in 1995-2004. Cellular automata (CA) were also used for the simulation of land use changes as a comparison. The study indicates that MAS has better performance for simulating complex cities than CA.