This study was designed (a) to test and (b) to compare proarrhythmic effects of levcromakalim and nicorandil; and (c) determine the mechanism of arrhythmia initiation by using high-resolution ventricular epicardial mapping on 44 Langendorff-perfused rabbit hearts. Eighteen hearts were kept intact and received incremental doses (1-500 microM) of levcromakalim, nicorandil, and isosorbide dinitrate. In 26 hearts, a thin layer of epicardium was obtained after endocardial cryotechnique (frozen hearts). In intact hearts, isosorbide dinitrate did not produce any arrhythmia. In contrast, levcromakalim induced spontaneous ventricular fibrillation (VF) in all hearts at 50 microM, whereas only one VF occurred at 500 microM nicorandil. These three drugs produced a dose-dependent bradycardia in intact hearts. In frozen hearts, arrhythmias were induced by 5 microM levcromakalim and 50 microM nicorandil. Isosorbide dinitrate had no proarrhythmogenic effect. Epicardial mapping showed that most of induced ventricular tachycardias were based on reentry around an arc of functional conduction block. Ventricular conduction velocities did not change, but levcromakalim and nicorandil shortened ventricular effective refractory period. We conclude that (a) levcromakalim and nicorandil, used in toxic concentrations, have direct proarrhythmic effects; (b) nicorandil proarrhythmogenic effects are 10 times less marked than those of levcromakalim (arrhythmia is solely the result of the potassium channel opener property of nicorandil); and (c) most of ventricular tachycardias induced are based on reentry.