The influence of the timing of a nontransmitted or transmitted atrial impulse on the atrioventricular (AV) conduction time of the subsequent impulse was studied in nine isolated rabbit hearts. AV conduction curves were determined by applying the atrial extrastimulus test. The extrastimulus was delivered preceded or not by an interposed atrial impulse whose coupling interval with respect to the last atrial beat of a basic train was kept constant at 100, 120, 140, 160, 175, 200, 225, 250, and 300 msec. In all experiments, there was a "concealment interval," i.e., the AV effective refractory period was longer than the atrial functional refractory period, and in seven experiments was comprised between 100 and 160 msec. For any given extrastimulus coupling interval in the presence of an interposed nontransmitted atrial impulse, AV conduction time was significantly greater than in its absence; the increase was greater than the longer the nontransmitted atrial impulse coupling interval, i.e., the shorter the subsequent transmitted impulse coupling interval with respect to the previous interposed nontransmitted impulse. The AV conduction curves relating the extrastimulus AV conduction time to its coupling interval with respect to the last atrial impulse of the basic train fitted to a hyperbolic model both in the absence of the interposed atrial impulse and in its presence (mean square residual 31 +/- 23 msec2), and the interposed atrial impulse modified the constants of the functions; the slope of the linear transformations was progressively more negative as the interposed atrial impulse was delayed. Furthermore, the effects of the interposed atrial impulse--transmitted or not--on AV conduction time of the subsequent impulse were qualitatively similar, their magnitude depending on the time elapsed were qualitatively similar, their magnitude depending on the time elapsed between the two.