Drug withdrawals over recent decades have triggered changes in the way that drug targets and screening programs are researched and designed. In the cases having the greatest impact, the reason for withdrawal was the reversible interaction of a drug or its metabolite with a single receptor, ion channel or enzyme (primary or secondary pharmacology). Once this interaction is identified, screens can be established and validated. When the mechanism is complex (eg, organ toxicity), however, such screens are difficult to implement and usually examine only the initial step, leading to considerable problems in extrapolation and risk definition. This review classifies drugs withdrawn from the US market over the last 25 years by their reasons for withdrawal, and examines how drug discovery programs have been modified in response to these events.