Drug registries are implemented after the authorization of new products and represent a tool for systematic collection of data aimed at obtaining additional knowledge on appropriateness, effectiveness and safety. The design of registries needs to be coherent with the main objective and a study protocol is required before the implementation. A registry aimed at the appropriateness of drug use should be primarily considered for high cost drugs when there is a risk, either for the patients' safety or for public expenditure, in using the drug outside the approved indications. Since the registry is a condition for the access to drugs, and all users are included, an extremely simplified data collection is required. However, the data should be available at regional level to allow record linkage procedures with other databases for conducting outcome studies. When registries are aimed at acquiring new information on the risk profile, the duration and the regional extension of data collection should be coherent with the expected incidence of events of interest. A great attention should be devoted in preventing that patients are lost to follow-up, since the reasons for being lost are frequently associated with harmful outcomes, such as adverse drug reactions. In a registry focused on effectiveness, the main aim consists in ascertaining the reasons (the prognostic factors), for possible discrepancies between premarketing studies and clinical practice. Taking into account the greater incidence of the expected events, there are fewer reasons for extending data collection to all users, whereas the main attention should focus on quality controls and the ascertainment of confounding factors. Given the relevance of the validity issues, in the set out of a registry it is important to think about ad hoc resources and the adequacy of infrastructures. As for any epidemiological study, an adequate qualification of the researcher/clinician in charge of conducting a registry should be guaranteed, together with independence in data analysis and freedom to publish all findings.