The efficacy, electrophysiologic effects, and side effects of adenosine triphosphate (ATP) and verapamil in the short-term management of paroxysmal reciprocating junctional tachycardia (PRJT) were compared in 20 patients. All patients had incuble sustained PRJT during control electrophysiologic study. Fourteen patients had PRJT involving a retrograde accessory pathway, and six patients had atrioventricular (AV) nodal reentrant tachycardia ('slow-fast' type). ATP, which has a very short half-life, was first administered (10 mg iv over 1 sec) during sustained PRJT. If PRJT did not terminate within 2 min, a bolus of 20 mg ATP was given. Verapamil (5 mg iv over 15 sec) was subsequently administered during sustained PRJT, and if latter did not terminate within 3 min another bolus of 5 mg verapamil was given. The cycle lengths of PRJT before administration of 10 or 20 mg ATP and 5 mg verapamil were similar. The 10 mg dose of ATP terminated PRJT in 17 of the 20 patients, and 20 mg ATP was required to terminate PRJT in the three remaining patients. The 5 mg dose of verapamil terminated PRJT in 15 patients, whereas an additional bolus of 5 mg terminated PRJT in one of the remaining five patients. The overall efficacy of ATP (20/20, 100%) was significantly greater than that of verapamil (16/20, 80%) (p < .05); however, there was no significant difference between the conversion rate of PRJT after administration of 10 mg ATP (17/20) and 5 mg verapamil (15/20). ATP terminated PRJT more quickly than verapamil (mean 24 sec vs mean 51 sec; p < .01). Termination of PRJT by either ATP or verapamil was mainly related to a block in the AV node in patients with accessory pathways and to a block in the antegrade slow pathway in patients with AV nodal reentry. Cycle length alternans before termination of tachycardia was observed more frequently after verapamil than after ATP (7/16 vs 1/20; p < .01). The total incidence of transient second-degree AV nodal block and various cardiac supraventricular and ventricular arrhythmias was higher after termination of PRJT by ATP than after verapamil (17/20 vs 5/16; p < .001). A higher incidence of transient but frequently uncomfortable noncardiac side effects was observed after ATP. We conclude that ATP (10 to 20 mg) is more effective and more rapid than verapamil (5 or 5 + 5 mg) in terminating PRJT but results in a higher incidence of cardiac and noncardiac side effects. We therefore recommend its use only if a cumulative dose of 10 mg verapamil has failed to terminate PRJT, if verapamil is contraindicated, or if a very rapid conversion to sinus rhythm is deemed urgent.