BACKGROUND The high-frequency contribution (i.e. 60-120 Hz) within the QRS complex of the surface ECG is supposed to be related to the arrhythmogenic substrate underlying sustained ventricular tachycardia in patients with coronary artery disease. One of the factors that prevents a wider clinical application of spectral analysis based on the fast Fourier transform is the considerable intra-individual variability of its results. This study presents a novel alternative approach of frequency-domain analysis, aimed at minimizing this methodological drawback. METHODS The proposed method uses gliding-window fast Fourier transform analysis of the signal-averaged ECG to provide values expressing the area ratio for two predefined frequency ranges (60-120 Hz divided by 0-120 Hz). The term gliding window is derived from the methodological principle by which the window analyzed is moved from the onset of the QRS complex into the ST segment. Values of area ratios obtained for every time instant of the interval of gliding are the basis for construction of curves. These are further analyzed in order to find patterns that characterize abnormal ventricular activation, with special emphasis on the identification of arrhythmogenic correlates. RESULTS In a group of 30 healthy subjects, the distribution of area ratio peaks was characterized according to their value and timing. The distribution of area ratio peaks in 43 patients with myocardial infarction differed substantially from the normal distribution in the control group. The amplitude of the peaks and their timing was not related to the occurrence of late ventricular potentials. CONCLUSIONS The method of gliding-window fast Fourier transform analysis eliminates the variability of results obtained by traditional spectral analysis of the ECG signal. A higher number and a different distribution of high-frequency peaks during the QRS complex in postmyocardial infarction patients reflect abnormal ventricular activation. However, late ventricular potentials are not related to a higher proportion of high frequencies in the corresponding time interval within the terminal QRS complex.