Case-control and cohort studies are the two principal types of observational studies. Case-control studies compare a group of patients with a given disease and a group of subjects without it (controls). An association between the risk factor tested (which may be a drug) and the disease is assessed by comparing the frequency of exposure to this risk factor in each group. In a cohort study, a large group of subjects initially free from the disease is followed over time. Exposure is assessed at the beginning of the study. The risk factor is associated with the disease if the incidence of the disease during follow-up is higher among exposed than non-exposed subjects. Associations observed in observational studies cannot be considered causal, in contrast to those in randomized controlled trials. Observational studies generate hypotheses that must be formally tested in randomized trials, even when all the elements supporting causality (dose-effect relation, biological plausibility, strength and independence of the association) are present. In trials, randomization, blinding and intention-to-treat analysis allow assessment of initial and secondary comparability between the two groups, so that any difference between the two groups can be attributed to the intervention or agent tested.