This paper addresses the crucial problem of changes and evolution in the context of database systems. We argue that the importance of change depends primarily on its nature. The nature of change impacts one or several phases of the database life-cycle. We propose a typology of changes based on three dimensions, namely the nature, the significance, and the time frame. These dimensions enable us to characterize the changes and to propose a set of evolution techniques facilitating the propagation of change in the database system. A framework for database evolution is presented. More generally, the framework provides a first step in the development of a new generation of guidance tool exploiting the potential of evolution techniques such as reverse engineering, forward engineer-ing, schema integration, and data integration. Our aim is to provide maintenance teams with realistic experiential database evolution environments. Finally, the implications for practice and for further research are discussed.