A combination of methods based on hardware, software and mathematical support, was used to filter an ECG signal, transmitted telemetrically to the computer. Clusterization of P-Q and R-R intervals was used as primary informative ECG signs, providing the basis for the diagnosis of the type of heart rhythm disorder. The comparison was made by correlation of ranges. An algorithm, based on unconventional clinical signs, was developed. An analysis of 672 electrocardiograms has demonstrated that mean sensitivity of the proposed automated diagnosis of the basic heart rhythm is 96.9%, and its specificity is 98.0%.