At supratherapeutic doses (2- to 5-fold), the T-type Ca²⁺antagonist mibefradil modifies the T/U wave of the human ECG. In this study, we show that this effect is observed in conscious monkeys and is duplicated by verapamil or diltiazem. We then evaluate the proarrhythmic risk of such alterations of cardiac repolarization by examining the actions of mibefradil on cardiac action potentials (APs). In isolated cardiomyocytes from guinea pigs or humans, mibefradil dose dependently shortens the plateau of the AP; this effect is similar to other Ca²⁺ antagonists and opposite to drugs having class III antiarrhythmic properties. The metabolites of mibefradil, singly or in combination, also shorten APs. In isolated rabbit hearts, noncardiodepressant concentrations of mibefradil have no effect on monophasic action potentials (MAPs), whereas cardiodepressant levels produce a slight nonsignificant lengthening. In hearts of open-chest bradycardic dogs, mibefradil has no effect on MAP dispersion or on QT interval and shortens MAPs slightly; although high doses produce atrioventricular block, likely through Ca²⁺ antagonism, arrhythmias are never observed. In contrast, d-sotalol lengthens QT interval and MAPs, increases dispersion, and produces arrhythmias. Together, these in vitro and in vivo results suggest that mibefradil carries no proarrhythmic risk despite changes in T/U wave morphology. Although these changes resemble those observed with class III compounds, they also are seen with nonproarrhythmic compounds such as verapamil and diltiazem. In conclusion, the classical models used in the present study could not link the changes in T/U wave morphology produced by mibefradil and verapamil to any experimental marker of proarrhythmic liability.