In this paper we present the development and application of a spatial model, named GEOMOD, for simulating land-use mid land-cover changes and their impacts on the carbon exchange between terrestrial ecosystems and the atmosphere. We first discuss the philosophy and principles on which the model is based. Then we describe in detail the model algorithm and structure. The model is tested using data from Peninsular Malaysia and Chiang Mai Province, Thailand, and applied to the entire region of South and Southeast Asia. The results show that the model is able to produce a satisfactory prediction over large temporal and spatial scale when based only on a few driving variables for pattern change. The change of carbon content and its geographical distribution is also simulated. While further refinement is needed to increase the prediction accuracy of the model, this study demonstrates the merit of the rule-driven approach in simulating systems with large spatial and temporal scales of which data are often inadequate for statistical approach.