We describe our efforts in providing REINAS (Realtime Environmental Information Network and Analysis System) weather visualization products on demand through the web. As the REINAS research enterprise transition into operational mode, there is increasing demand for wide and effective distribution of results. To meet this demand, we provide weather products on the web as both images and VRML worlds. The viewer selects geographicextents and time range, as well as specific weatherproducts. Data is then pulled out of the database based on the user’s specifications and fed to the visualization tools to produce the image or VRML file. This paper describes the architecture of this end-to-end system for delivering sensor and forecast data through the database and to the public as VRML-based visualization products. Key words and phrases: VRML, HDF, database,environmental visualization, weather, data fusion, data assimilation.