Health registries and observational studies are different entities, and although several registries are conceived as observational cohort studies, or provide the data that are needed for cohort or case-control studies, this is not always true: many registries have specific aims that imply methodological standards that differ completely from those of an observational study. Of course, whenever a Registry is conceived with the aim of conducting an observational study, this will dictate its methods. The difference between a registry and the simple assembling and analysis of routine clinical data must be stressed as well: a registry may require that collection of data from different administrative sources, or of original data not available elsewhere, and must guarantee the quality and completeness of the information it contains. It has been repeatedly suggested that the huge databases of health information that are currently assembled by most health governing bodies should be exploited directly for evaluations of appropriateness and effectiveness. This is extremely dangerous, because the inherent potential biases already affecting all observational studies may completely undermine their validity if the currently accepted methodological standards are relaxed. Similarly, the use of observational studies for the assessment of the effectiveness of health technologies must be considered with caution.