Our objective was to determine whether neuronally induced atrial arrhythmias can be modified by -adrenergic receptor blockade. In 30 anaesthetized dogs, trains of 5 electrical stimuli (1 mA; 1 ms) were delivered immediately after the P wave of the ECG to mediastinal nerves associated with the superior vena cava. Regional atrial electrical events were monitored with 191 atrial unipolar electrodes. Mediastinal nerve sites were identified that reproducibly initiated atrial arrhythmias. These sites were then re-stimulated following: 1h (time-control, n=6), or the i.v. administration of naftopidil (1-adrenergic blocker: 0.2 mg/kg, n=6), yohimbine (2-adrenergic blocker: 1 mg/kg, n=6) or both (n=8). A ganglionic blocker (hexamethonium: 1 mg/kg) was tested in 4 dogs. Stimulation of mediastinal nerves sites consistently elicited atrial tachyarrhythmias. Repeat stimulation after 1 h in the time-control group exerted a 19% decrease of the sites still able to induce atrial tachyarrhythmias. Hexamethonium inactivated 78% of the previously active sites. Combined -adrenoceptor blockade inactivated 72% of the previously active sites. Bradycardia responses, induced by mediastinal nerve stimulation, were blunted by hexamethonium but not by 1,2-adrenergic blockade. Naftopidil or yohimbine alone eliminated atrial arrhythmia induction from 31% and 34% of the sites (similar to time control). We conclude that heterogeneous activation of the intrinsic cardiac nervous system results in atrial arrhythmias that involve intrinsic cardiac neuronal -adrenoceptors. In contrast to the global suppression exerted by hexamethonium, we conclude that -adrenoceptor blockade targets intrinsic cardiac local circuit neurons involved in arrhythmia formation and not the flow-through efferent projections of the cardiac nervous system.