The premature atrial stimulation technique was investigated as a method of measuring sinoatrial conduction time in the rabbit. Fifteen studies were performed in which intracellular recordings were obtained from a sinus node cell and atrial electrical activity was recorded from the crista terminalis by a surface electrogram. An additional ten studies were performed without microelectrode recordings. Atrial premature depolarizations late in the cardiac cycle produced compensatory atrial return cycles, but earlier premature depolarizations produced less than compensatory return cycles. Compensatory return cycles only occurred with atrial premature depolarizations that failed to capture the sinus node cell. The transition from compensatory to less than compensatory return cycles occurred with late atrial premature depolarizations that failed to capture the sinus node. Therefore, the transition from compensatory to less than compensatory atrial return cycles failed to indicate sinus node capture by an atrial premature depolarization. Although these premature depolarizations were too late to capture the sinus node cell, they still shortened the sinus node and atrial return cycles to make the atrial return cycle less than compensatory. This shortening of the sinus node return cycle was due to a shortening of the sinus node action potential by electrotonic interaction between sinus node and adjacent cells during repolarization. The electrotonic effect resulted in shortening of the sinus node action potential and accounted for the poor correlation (?• = 0.64) between estimated and measured values of sinoatrial conduction time. These data indicate that there are significant limitations to the use of the premature atrial stimulation technique for estimating sinoatrial conduction time.