Routine healthcare data is becoming widely available, usually as a result of administrative systems. Other related data are also often available, such as biochemistry results, mortality data, and sometimes prescribing data. These records are often linked via a common identification system or by probability matching techniques. These data sources offer many opportunities to undertake research, and where prescription data are recorded and linked, the facility to research the outcome of drug use often exists. There are now a number of research agencies around the world that use these large routine data sources to undertake drug safety and outcome studies. The purpose of this commentary is to describe some of the history behind the development of these systems, illustrate some of their uses with respect to postmarketing drug safety and to other healthcare research objectives. The review then describes the data sources necessary to develop a system that would offer an optimal system to undertake a range of studies, including population drug safety surveillance. There are both positive and negative considerations when using routine data. On the positive side, these data come from 'real life' experiences and not from the clinical trial situation. On the other hand, there are important biases to be aware of such as confounding by indication. On the whole, it is argued that large databases originating from routine healthcare procedures have an important role to play in the cost-effective prescription drug use in the postmarketing setting. These systems cannot replace other methods of drug safety evaluation but they do offer an important adjunct to spontaneous reporting systems.