In order to create and evaluate a model sensitive to QT-dependent proarrhythmic effects of drugs, a long QT syndrome was produced in chronically instrumented dogs with bradycardia and hypokalemia. Bradycardia (mean cycle length: 1495 +/- 78 msec) was provided by permanent atrioventricular block and hypokalemia (K+ = 2.6 +/- 0.05 mmol/l) by high doses of diuretics. To evaluate that model, six of these conscious dogs were subjected to quinidine, flecainide, lidocaine, propranolol and sotalol infusions. In crossover design, drugs were infused i.v. at rates allowing stable and nontoxic drug plasma levels during the experiment. Four-lead ECGs were recorded for arrhythmias for 30 min before (base line) and 75 min after onset of infusion. Ventricular cycle length was increased dramatically by sotalol, lidocaine and propranolol (+618 +/- 192, +388 +/- 125 and +329 +/- 114 msec, respectively) and QT interval was increased by sotalol, quinidine and flecainide (+56 +/- 8, +31 +/- 7.9 and +20 +/- 5.7 msec, respectively). Quinidine and sotalol, but not flecainide, propranolol or lidocaine, exhibited significant arrhythmogenic activities. During quinidine infusion, most dogs exhibited some ventricular arrhythmias whose most severe forms were runs of ventricular tachycardia. These arrhythmias were suppressed by pacing at high rates. During sotalol infusion, five out of six dogs exhibited typical "torsades de pointes." This incidence was not related to the slowing effects of sotalol on idioventricular pacemakers, because a similar incidence was obtained in five complementary dogs paced at 40 bpm. It could be related to dose, because torsades de pointes occurred only once in another group of five dogs receiving half the dose used in the controlled study. Only quinidine and sotalol, but not propranolol, flecainide or lidocaine, are clinically associated to torsades de pointes. They were also the only drugs associated with proarrhythmic events in the present study, a fact suggesting that QT-dependent arrhythmogenic effects of drugs can be reliably evaluated in conscious hypokalemic dogs with complete atrioventricular block.