Based on the MapGuide module of WebGIS and .net technology, the expert system for prediction and forecast of agricultural pests on the Web was designed and developed. The pest categories were bio-regionalized in the system. The experts’ knowledge, abstracted from their experiences and including the description of the knowledge, the critical values for the description, terms for distinguishes, and the grade of pests harm, formed a Web model and was stored in tables supported by SQL Server 2000. Expert-knowledge-based forward reasoning and case-based reasoning (CBR) were employed in the inference engine. Users can provide the necessary field-monitoring data to the system with the wizard and the guide of hint in the interface, and can get the potential grade of pests harm in GIS thematic map, chart and text. The demonstrations show that the system is easy to use and the prediction is accurate and intuitive.