To give some explanation for atrial malsensing in dual chamber pacing that occurs only during exercise, atrial electrograms from 33 patients were telemetrically recorded and analyzed in both the time and frequency domains. During exercise, an overall decrease from 6.4 +/- 1.9 to 5.6 +/- 1.9 mV (-11%) in the atrial signal amplitude was noted. Despite considerable variability among patients, marked changes occurred in 15 patients whose signals diminished by 11 to 49%. Slew rates showed a similar decrease from 1.35 +/- 0.45 to 1.18 +/- 0.45 V/s (-10.8%), with individual changes of as much as -51%. Signal attenuation in the time domain correlated well with frequency data, exhibiting a highly significant reduction of signal energy between 25 and 105 Hz. However, spectral distribution changed from rest to exercise, with a relative increase of signal energy in the range between 5 and 25 Hz and a decrease at higher frequencies. Individual changes differed widely when low (15 to 65 Hz) and high (65 to 115 Hz) frequencies were compared, but in a group of 11 patients signal attenuation in the high frequency band was more pronounced (-45%) than in the low frequency band (-23%). The clinical impact of the change in frequency distribution during ergometry was visualized by computer simulation of two different (low and high bandpass) filters. Although in individual patients, both characteristics may be favorable with respect to atrial sensing, it was observed in 11 patients that high pass filtering attenuates signal amplitudes by 10 to 24% in excess of the variation without filtering.(ABSTRACT TRUNCATED AT 250 WORDS)