Previous research on rice productivity potential models, based on a single point in space and then extrapolated, is not ideal and is difficult to scale up. This study designed a rice productivity potential forecast system,which makes comprehensive use of spatial data and attribute data by integrating the model with geographic information systems according to the characteristics of rice productivity potential model and raster computing function. Following the idea of the software engineering and using COM technology, ArcGIS Engine components were embedded in Visual Studio.NET 2005 development environment. This system can achieve the goals of more systematic model parameters, more efficient calculation and more accurate analysis. This system can be used to predict potential and dynamic rice production. It is also important for the management of agricultural resources.