We used cesium chloride (CsCl) for electrophysiologic studies in canine hearts in vivo and in vitro to examine the mechanisms underlying ventricular arrhythmias that are related to prolonged repolarization. Cesium is known to depress normal ventricular automaticity and some experimental arrhythmias by blocking delayed outward currents and prolonging action potential duration. In 10 dogs in normal sinus rhythm, 1 to 1.5 mM/kg iv CsCl prolonged the QT (QU) interval and induced ventricular ectopy in all, including multiform ventricular tachycardia. In 12 dogs with atrioventricular block, 1 to 1.5 mM/kg iv CsCl produced marked suppression of idioventricular rates (from 45 +/- 6 to 8 +/- 4 beats/min). These low rates were then associated with bigeminy or bursts of multiform ventricular arrhythmia. Pacing at rates of 60 beats/min or more suppressed these arrhythmias. Low doses of tetrodotoxin (1 microgram/kg) also abolished these bradycardia-dependent arrhythmias without affecting the amplitude of ventricular electrograms. Tissue concentrations of cesium were determined by anatomic absorption spectroscopy in five dogs after injection of 1 mM/kg CsCl. Thirty minutes after the injection, cesium levels in Purkinje fibers were 5.3 +/- 1.0 mM/kg, levels in ventricular muscle were 4.6 +/- 0.9 mM/kg, and levels in atrial muscle were 4.1 +/- 0.8 mM/kg. In eight isolated endocardial preparations from canine ventricles, standard microelectrode techniques were used to study the effects of superfusion with 5 mM cesium. After 30 min, we observed early afterdepolarizations interrupting phase 3 of Purkinje fiber action potentials that already showed prolonged repolarization. Slowing the rate generated single or multiple action potentials arising from partially repolarized levels of membrane potentials (-80 to -65 mV). Pacing rates of 30 to 60 beats/min diminished the afterdepolarizations and suppressed the spontaneous beats. Tetrodotoxin at a concentration of 10(-8) g/ml, which did not affect upstroke velocity, abolished the afterpotentials. We conclude that cesium induced bradycardia-dependent ventricular arrhythmias caused by early afterdepolarizations. These data suggest that an inward current, probably carried by sodium ions, appears to be essential for the occurrence of this phenomenon. The association of delayed repolarization, afterdepolarizations, and triggered activity has similarities to the phenomenon of drug-induced prolongation of the QTU interval associated with multiform ventricular tachycardia in humans, i.e. "torsades de pointes."