This paper gives a survey on the principles, methodology, and clinical applications of the signal-averaging technique. Bioelectric signals of the heart at the microvolt level are overwhelmed by basal noise in high-gain amplified EKG registrations. By summation and averaging of several hundreds of cardiac cycles, basal noise can be minimized and the signal of interest accumulates. This improvement of the signal-noise ratio primarily depends on the number of cycles averaged, e.g., by 10:1 when 100 cycles are averaged (approximately, the square root of the number of averaged cycles). At present, signal-averaging computers in use are rather variable with respect to data acquisition, filter settings, trigger mechanism/stability, and data processing. These properties have to be taken into account when the results of different groups using different types of averaging computers have to be compared. At present, several applications of the signal-averaging technique in clinical cardiology have been proposed: (1) Surface EKG--registrations of pre-P potentials, of His bundle potentials, and of ventricular late potentials, as well as analysis of the frequency spectrum of the QRS complex in patients with myocardial infarction; (2) intracardiac EKG--recordings of pre-P potentials and of A-V nodal potentials; (3) surface phonocardiogram--detection of low-amplitude, late diastolic signals; (4) Intracardiac pressure analysis--recording of average phasic pressure changes. The most promising and clinically proven application of the signal-averaging technique, at present, seems to be the registration of His bundle potentials as well as of ventricular late potentials from the body surface. Future efforts will need to be made to record cardiac microvolt potentials beat by beat from the body surface by means of specially designed EKG recorders.