The concept of maintaining a medical data dictionary as a HIS core component was fundamental for all HIS development phases since the mid eighties at Giessen University Hospital. Being influenced by an early experimental installation of the HELP hospital information system and its PTXT data dictionary, we kept this approach through a number of development cycles of our own hospital information system. While our first data dictionary implementation (GMDD) was still very close to the PTXT structure (polyhierarchical design with an eight level hierarchy), the second generation dictionary (MDD-GIPHARM) has already been designed using a more flexible semantic network model. GMDD was a mainframe development (realized on Tandem Computers) based on the Tandem Nonstop SQL RDBMS. The major clinical applications established on top of the GMDD were laboratory results review, diagnosis documentation and physician discharge summaries. The MDD-GIPHARM development was initiated on PC-basis as the core of a rheumatology departmental system using MS-Access and then further enhanced within a research project to build knowledge-based functions for drug therapy. A first set of such functions based on MDD-GIPHARM is in routine use since 1996. Our current focus is to enhance MDD-GIPHARM towards an application independent vocabulary server (GDDS), which may be used for a variety of applications with the intranet of Giessen University Hospital. In this paper the evolutionary development of those data dictionary concepts at Giessen University Hospital is illustrated and compared with international activities in the last decade.