Fire is used in the management of ecosystems worldwide because it is a relatively inexpensive means of manipulating thousands of hectares of vegetation. Deciding how, where and when to apply fire depends primarily on the management objectives of the area concerned. The decision to ignite vegetation is generally subjective and depends on the experience of the fire manager. To facilitate this process, ancillary tools, forming a decision support system, need to be constructed. In this study a spatial model has been developed that is capable of simulating fire dynamics in savanna ecosystems. The fire growth model integrates spatial fuel and topographic data with temporal weather, wind settings and fuel moistures to produce a time-evolving fire front. Spatial information required to operate the model was obtained through remote sensing techniques, using Landsat Thematic Mapper (TM) satellite imagery, and existing Geographic Information Systems (GIS) coverage's. Implementation of the simulation model to hypothetical landscapes under various scenarios of fuel, weather and topography produced fire fronts that were found to be in good agreement with experience of observed fires. The model was applied actual fire events using information for prescribed burning operations conducted in Mkuze Game Reserve during 1997. Predicted fire fronts were found to accurately resemble the observed fire boundaries in all simulations. / Thesis (M.Sc.)-University of Natal, Pietermaritzburg, 1999.