Automatic adaptation of the atrial sensitivity was evaluated in 18 patients with dual chamber pacemakers (Intermedics, Inc., Relay) in the unipolar mode. After atrial sensitivity was stabilized in the upright position, patients underwent a 1.0 W/kg body weight exercise for 5 minutes. A 24-hour Holter ECG was recorded, and the maximum and minimum atrial sensitivity values reached were stored in the memory of the pulse generator. In a second series of 12 patients, Holter ECGs were recorded twice, starting with the same sensitivity but with automatic adaptation alternately switched "on" or "off." Results of the exercise test: mean atrial sensitivity declined from 2.30 +/- 0.77 mV to 2.03 +/- 0.68 mV. There was no change in five patients, a slight increase in two patients, and lowering of the atrial sensitivity was observed in 11 patients, the difference ranging from 0.2 to 1.0 mV. A total of two P waves in two patients were missed by the atrial amplifier. The minimum and maximum sensitivity reached during Holter monitoring averaged 2.31 +/- 0.67 mV versus 1.72 +/- 0.71 mV (difference 0-1.7 mV). Normal pacemaker function was found in six patients, including one patient without any intrinsic atrial activity. Malsensing of less than five P waves occurred in four patients. More than 50 sensing defects resulted from ectopic atrial beats (four patients). We observed atrial oversensing in three cases; one patient showed atrial over- and undersensing. The comparison between fixed and variable sensitivity did not reveal any superiority of automatic adaptation.(ABSTRACT TRUNCATED AT 250 WORDS)