This paper presents a flexible system, DIVE-ON, for the purpose of visual data mining. A new approach to interactively visualize and explore N-dimensional data warehouses in an immersed virtual environment is put forth. DIVE-ON is capable of constructing a multidimensional data model on a remote system, transporting pertinent views to a CAVE, creating an immersed virtual environment and providing an interactive data mining toolset. DIVE-ON architecture emphasizes the development of two independent subsystems, a visualization environment and a virtual data warehouse. The first objective of our research is to examine the possibility of effective mining, and manipulating views with little or no instructional help by providing an environment that is built around the human’s visual, sensorimotor, and spatial knowledge acquisition abilities. The second goal is to create a highly transparent and centralized data warehouse that integrates various distributed data sources. Within the warehouse, DIVE-ON incorporates an XML-based multidimensional query language (XMDQL) to circulate the queries among the distributed data sources.