Some drugs have been reported to induce severe ventricular arrhythmias, including torsades de pointes, and have been responsible in some cases for sudden death of patients. Although the mechanisms of these arrhythmias are not well understood, they are often, but not always, associated with QT interval prolongation. Regulatory authorities (CPMP in Europe) have recently pointed out the necessity to assess most carefully the potential, especially of non-cardiovascular drugs, for QT interval prolongation. Different methodological approaches are presented in this paper and experimental protocols are suggested; limitations and advantages of the presently available in vitro and in vivo models are discussed. It appears that both in vitro and in vivo approaches are complementary. In particular it is pointed out that only the in vitro models using isolated cardiac tissues (Purkinje fibres or papillary muscles) enable assessment of the drug properties under low cardiac rhythm conditions. This model allows us to mimic pathological situations of long QT interval (such as acquired or congenital long QT syndrome) in which most of the major clinical problems are encountered. Finally, a strategy for the preclinical assessment of the potential of a molecule for QT interval prolongation is presented.