Due to the evolution of social needs, together with new potentialities in information and communication technology, new considerations are emerging, so the goal of this paper will be to emphasize the new directions of databases for GIS. The use of GPS, together with mobility leads to new location-based services (LBS). In database terms, this means that the information must be pervasive, i.e. accessible everywhere by means of antennas and special server architecture. Among the critical issues, let us mention mobile queries and transactions which imply the design of high performance caching systems. Another evolution regards real time GIS, not only for LBS, but also for risks and disaster management. The key-points are the existence of myriads of sensors regularly sending measures to some monitoring centre. Here the technological barriers are essentially the performances of existing spatio-temporal databases systems which cannot correctly handled real time data. The design of real time GIS is clearly a new technological barriers. Applications such as meteorology deal with data coming from continuous phenomena, known as field oriented databases. Presently, no marketed database systems allow a continuous vision of the reality from a user’s point of view. Cadastral applications, maintenance and geology implies the necessity of having real 3D GIS, and as a consequence database systems able to store query and visualize 3D objects. NICT implies also the reorganization of the processes, especially for urban and environmental planning. So, new functionalities must be integrated allowing cooperation (groupware) between actors and participation of the citizens. Moreover, since decision-making procedures are more and more backed on computers and database contents, spatial analyses techniques must be based on data with a very high quality level. So, geographic databases must encompass efficient quality control mechanisms in order to reduce errors in decision making. The importance of Internet is pushing the design of GIS able to work not only at server level, but also at client level. The consequence is not only geographic data structure, but also the existence of cartographic tools at client level. Finally, another important evolution is the use of XML, not only for storing geographic data, but also metadata.