[Medical registries. Benefits and limitations]

Traditionally, registers are understood as tools of epidemiological research. Recently, medical registers have encountered new tasks in quality management and research on medical care. When registers are complete and representative for a target population, they complement the epidemiological and clinical studies that usually focus on specific endpoints using selected study populations. In this process, registers can benefit substantially from the established methodology of planned studies, for instance by application of survey sampling techniques. In order to fulfil the steadily rising expectations, registers have to meet increased methodological standards. Organization and data management should be orientated towards the standards that are meanwhile generally accepted for randomized controlled trials. The evaluation has to go beyond simple descriptive statistics towards complex statistical model building in order to cope with the unavoidable problem of internal validity of comparisons within registers. If registers fulfil these requirements, they are valuable sources of information (that cannot be replaced by other tools) on the spread of risk factors and diseases and on the distribution and usefulness of preventive measures and therapies; general proof of their efficacy should be provided by randomized clinical trials.