Summary: In 70 patients (3 females and 67 males), aged 16-72 years (mean: 51 +9 years), the low noise ECG was recorded from body surface by the signal averaging and the high resolution beat-to-beat techniques. We found 61 patients were suffering from coronary heart disease, 4 had atypical coronary heart disease (syndrome X), 4 had dilatative cardiomyopathy, and one had the long QT syndrome (Romano-Ward syndrome). We found the following recovery rates for ventricular late potentials within the ST segment with the averaging technique: clearcut in 13/53 patients, doubtful in 16/53 patients, and late potentials absent in 26/53 patients. With the beat-to-beat technique the following recovery rates were found: clearcut late potentials in 27/70 patients, doubtful in 23/70 patients, none in 20/70 patients, and intermittently occumng late potentials in 18/70 patients (categorized as doubtful late potentials). When comparing the detection of late potentials with both methods in individual patients, we found concordant results in 39/53 patients studied (positive with both methods in 24/53 patients, negative with both methods in 15/53 patients), and discordant results in 14/53 patients (positive with the beat-to-beat technique and negative with the averaging technique in 12 individuals, negative with the beat-to-beat technique and positive with the averaging technique in the remaining 2 patients). The correlation between the incidence of late potentials and the presence of exercise-induced myocardial ischemia (submaximal bicycle exercise) was higher when using the high resolution beat-to-beat technique, as holds also true for the correla- tion to complex ventricular arrhythmias. Dynamic changes of late potentials were only detectable by the beat-to-beat technique, which seems to be a considerable advantage over the averaging technique. In the clinical setting, both methods for retrieval of ventricular late potentials should be applied in order to increase the information and to provide the facilities of investigating larger patient populations for screening purposes.