When a clinician institutes a therapy, he/she has to consider, besides its efficacy, the harm a substance that is foreign to the body may cause. Nowadays, with the number of drugs available, more and more iatrogenic events are occurring, to the extent that the term 'iatroepidemic' has been coined to describe the phenomenon. Harm can be assessed by experimental studies, but this design raises ethical as well as financial issues, so the best evidence is gathered from observational data, mainly case-control and cohort studies. In previous articles we discussed the methodology to evaluate therapeutic benefit. In this article we aim to review designs for determining harm.