Lifelong electronic health records can supply valuable information for research, quality management and health politics in addition to supporting treatment of patients. Based on experiences with scientific data analysis in a university hospital environment, requirements on cross-institutional analysis of electronic health records in a healthcare system are discussed. The concept of archetypes can play a key role in this context. Archetypes can be utilized in data analysis for visualization, semantic linkage and finally for standardized data transfer. Keywords. Archival-repository systems for medical records-EPR-CPR-EMR, Standards, Data analysis-extraction tools, Epidemiological research Hospital IS