Benefit-risk assessment should be ongoing during the life cycle of a pharmaceutical agent. New products are subjected to rigorous registration laws and rules, which attempt to assure the availability and validity of evidence. For older products, bias in benefit-risk assessment is more likely, as a number of safeguards were not in place at the time these products were registered. This issue of bias in benefit-risk assessment of older products is illustrated here with an example: buflomedil in intermittent claudication. Data on efficacy were retrieved from a Cochrane systematic review. Data on safety were obtained by comparing the number of reports of serious adverse events and fatalities published in the literature with those reported in postmarketing surveillance databases. In the case of efficacy, the slim basis of evidence for the benefit of buflomedil is undermined by documented publication bias. In the case of safety, bias in reporting to international safety databases is illustrated by the discrepancy between the number of drug-related deaths published in the literature (20), the potentially drug-related deaths in the WHO database (20) and deaths attributed to buflomedil in the database of the international marketing authorization holder (11). In older products, efficacy cannot be evaluated without a thorough search for publication bias. For safety, case reporting of drug-related serious events and deaths in the literature remains a necessary instrument for risk appraisal of older medicines, despite the existence of postmarketing safety databases. The enforcement of efficient communication between healthcare workers, drug companies, national centres of pharmacovigilance, national poison centers and the WHO is necessary to ensure the validity of postmarketing surveillance reporting systems. Drugs considered obsolete because of unfavourable benefit-risk assessment should not be allowed to stay on the market.