The use of large databases has a high potential for future epidemiological research. Databases consisting of primary data as well as routine data that are mainly collected for administrative purposes will be of interest. This paper presents some large databases and some that are most relevant for epidemiological research, respectively, and shows how they might be exploited for research purposes. To make better use of large databases there are new requirements and challenges that we have to face regarding appropriate methods for their analysis. Some examples of such methods are given. An example of pharmaco-epidemiology illustrates linkage of several administrative databases and specific problems that have to be solved in this context. Finally, future challenges are discussed that have to be overcome on the way towards establishment, maintenance and analysis of large databases, especially regarding data protection, data quality and regular updates.