AIMS Life-threatening arrhythmias are a major problem in chronic heart failure (CHF). The aim of the present study was to investigate the mechanism underlying the low proarrhythmic potential of amiodarone in a model of pacing-induced heart failure. METHODS AND RESULTS Heart failure was induced in 35 female rabbits by rapid ventricular pacing, resulting in a significant decrease of ejection fraction. Thirty-four rabbits were sham-operated. In 17 of 35 CHF-rabbits and 20 of 34 'sham'-rabbits, amiodarone (50 mg/kg/day) was fed for 6 weeks. Eight monophasic action potentials and a simultaneously recorded 12-lead electrocardiogram showed prolongation of QT-interval and action potential duration (APD(90)) in all failing hearts (P< 0.05). Amiodarone pre-treatment led to a prolongation of APD(90) (+19 ms) as compared with sham-controlled hearts but showed only a marginal effect on APD(90) in failing hearts. Infusion of sotalol (50-100 µM) led to a significant prolongation of APD(90) in sham and a further prolongation of APD(90) in failing hearts [+55 ms (50 µM); +70 ms (100 µM); P< 0.01 as compared with sham hearts]. Sotalol led to a triangular action potential configuration in sham and failing hearts, whereas amiodarone did not cause triangularization but caused a rapid phase-III repolarization. Moreover, amiodarone did not increase dispersion of repolarization either in sham or in failing hearts. Infusion of sotalol led to a significant increase in dispersion of repolarization in sham (+29 ms) and a further increase in failing hearts (+67 ms; P< 0.05). CONCLUSION Chronic amiodarone results in a rapid phase-III-repolarization and does not increase dispersion of repolarization. These electrophysiological findings are present in healthy hearts and are preserved in heart failure. This contributes to the low proarrhythmic potential of amiodarone in heart failure.