Cardiac micropotentials are considered to have a predictive value in critical ventricular tachycardia or sudden death. These micropotentials are obtained by numeric filtration of the result of sequential averaging of about 200 systoles (i.e. of measurement at about 3 min interval) which is significantly influenced by known intraindividual ECG variability. It follows from our previous studies that the non-dipolar residue (i.e. the sum of all components of an equivalent source of the heart electrical field with the exception of the first three dominant dipolar components) corresponds by its nature to the cardiac micropotentials, i.e. to late potentials. Verification of this hypothesis utilizing singular value decomposition and replacing the sequential averaging by "surface" averaging of the matrix of synchronously measured ECGs is the aim of this project. The results of the present study can be considered as a confirmation of this hypothesis. These results provide a better understanding of the structure of the body surface potential distribution and for clinical purposes they make it possible to attain cardiac micropotentials (late potentials) from one systole.