Medicinal products that, as an unwanted effect, prolong the QT interval of the electrocardiogram (ECG) can trigger episodes of polymorphic ventricular dysrhythmias, called torsades de pointes, which occasionally culminate in sudden death. The accurate measurement of QT interval requires the adoption of appropriate criteria of recording, measurement and data processing. Traditionally, QT interval is standardised to a reference heart rate of 60 beats/min by using the Bazett algorithm. However, this correction method can bias observed QT intervals in either direction. The ECG reflects cardiac electrical currents generated by ions (Na+, K+ and Ca2+) entering and leaving the cytosol mainly via transmembrane channels. Na+ and Ca2+ carry inward depolarising currents (INa, ICa) whereas K+ carries outward repolarising currents (Ito, IKr, IKS and IK1). Sometimes, a prolonged QT interval is a desired drug effect but, more commonly it is not, and reflects abnormalities in cardiac repolarisation heralding torsades de pointes. Furthermore, the potential torsadogenic activity of drugs is favoured by concurrent cardiac risk factors (old age, female gender, bradycardia, electrolyte imbalances, cardiac diseases etc.) which reduce cardiac repolarisation reserve. The evaluation of the cardiac safety of drug candidates can be started by determining their potency as IKr blockers in cloned Human Ether-a-go-go Related Gene (HERG) channels expressed in mammalian cells. Compounds passing successfully this test (desirable cardiac safety index > 30, calculated as ratio of IC50 against IKr over ED50 determined in an efficacy test) should be further investigated in other relevant human cardiac ion currents, in in vitro animal heart preparations and finally in in vivo pharmacodynamic models. The decision as to whether the potential benefit of a new drug outweighs the cardiac risk inherent in its therapeutic use should be made in the light of the condition that it is expected to treat and with reference to alternative drug therapies. If a drug represents a unique therapeutic advance, non-clinical and clinical signals of unsatisfactory cardiac safety may not constitute sufficient grounds to abandon its development. However, if the drug offers only marginal benefits over existing therapies, decisions concerning its possible development should be taken by corporate policy makers.