The behavior of ventriculoatrial conduction (VAC) during exercise remains unknown. In order to determine its characteristics and the consequences it might have on dual chamber pacemaker technology and programming, 17 patients underwent an electrophysiological study (EPS) of atrioventricular conduction (AVC) and of VAC during a protocol including three steps: supine rest, upright position, and finally during cycloergometric exercise; the measurements were done at progressively increasing pacing rates. During a preimplantation EPS, Wenckebach points AVC-W and VAC-W and conduction times, AVCT and VACT (as a function of pacing rate), were measured in ten consecutive patients using temporary leads and an external device. In another study, AVCT, VACT, AVC-W, and VAC-W were measured by telemetric recordings under identical conditions in seven patients implanted earlier with a DDD pacemaker. A 1/1 VAC was observed in 7/17 patients (41%) at rest, and in 13/17 patients (76%) at the end of the protocol; VAC was never observed in 4/17 patients (23%), but occurred in six of the ten patients initially free, three standing at rest and three on exercise. For all patients, the VAC behavior remained of "nodal" type, indicated by a progressive increase in VACT as pacing rate rose up to the VAC-W point. Neither the existence of exercise-induced VAC nor the maximal VACT-W could be predicted from AVC or VAC data obtained at rest. However, at the same pacing rates, standing up and exercise induced a shortening effect on VACT, and improved the VAC-W by an average of 33%. These results suggest that the electrophysiological behavior of VAC does not obey any general rule and cannot be predicted individually. It would thus appear unwise to base pacemaker mediated tachycardia (PMT) protection solely on long postventricular atrial refractory period (PVARP) programming in DDD patients. This work also revealed the potential risks of a rate responsive auto-adaptive PVARP algorithm as proposed in certain new devices.