Multiple coordinated views have often been used for visual analytics purposes over the last years. In this context, if the exploration of 2D visualizations is not an obstacle, adding an extra dimension can be an issue. The interaction with multiple 3D visualizations in 2D conventional displays lacks usability and does not guarantee the usefulness the extra dimension would provide. Immersive visualization techniques can potentially fulfill these gaps by providing 3D visualizations and novel 3D interactions simultaneously. In this paper, we propose a new approach for interacting with composite and multiple coordinated visualizations in immersive virtual environments. We use a 3D-WIMP-like concept, i.e., virtual cubes (Spaces), for encapsulating views, which the user can freely control in the virtual environment. Moreover, operations like “cloning” and “coordinated interactions” features provide a way for performing composed tasks. We compared our approach with a desktop version to evaluate its performance when dealing with composed tasks. A user study with 19 participants was conducted, and the results show that the immersive approach has advantages over the corresponding desktop version regarding interaction with multiple coordinated 3D views.