Visualization is an important technique for understanding and exploring complex models. To be useful, visualizations have to be specifically tailored towards the visualization task and the analyzed model. Many standard charts or graph-based visualizations exist, but need to be mapped to the concepts of the model under study. Existing model visualization tools often have predetermined visualization kinds and content, impeding reuse of standard visualizations for various purposes, or lacking the ability to flexibly map source model concepts to different visualization elements. This paper presents ELVIZ ( Every Language Visualization”), an approach to vi- ” sualization, generic regarding both the source model, and the kind and content of the visualization. ELVIZ applies model-driven engineering techniques both to transform arbitrary source models into the desired visualization models, and to generate said model transformations from a query-based mapping of source model concepts to visualization concepts. This cleanly decouples source and visualization meta-models, allowing to reuse and combine standard visualizations for various source models. The ELVIZ approach is applied to scenarios from software visualization in software evolution and measuring energy consumption of mobile applications, using different kinds of visualizations.