Depending on the research problem, the epidemiological evaluation of drugs must be based on a combination of research strategies that include experimental methods (controlled trials), and non-experimental methods. Among the latter are correlational studies; spontaneous reporting systems and published case reports; cohort studies; and case-control studies. The spread of computer technology may, by record linkage, facilitate cohort or case-control studies. Comprehensive drug surveillance must be based on the combined use of each research strategy, as appropriate.