This study compared short-term (1 day) and long-term (1 week and 1 month) reproducibility of the numerical results of 21 individual indices of time-domain analysis, spectral temporal mapping, and spectral turbulence analysis of the signal-averaged electrocardiogram in 28 healthy volunteers (18 men, 10 women, mean age, 31 years; range, 22-39 years). For each of the 21 indices and for each of the repeated recordings (ie, recordings on day 1, 7, and 30), the mean relative error versus recording of day 0 was calculated together with the rank correlation coefficient. The study showed that (1) reproducibility of time-domain analysis with the 40-Hz high-pass filter was superior to that of time-domain analysis performed with the 25-Hz high-pass filter, (2) reproducibility of time-domain variables was similar to that of spectral turbulence analysis of the entire QRS complex and was superior to that of spectral temporal mapping, (3) reproducibility of frequency-domain techniques not based on time-domain measurements (such as spectral turbulence analysis of the entire QRS complex) was significantly better than that of frequency-domain techniques combined with time-domain measurements (such as spectral turbulence analysis of the low-power terminal QRS region or spectral temporal mapping), and (4) short- and long-term reproducibility of signal-averaged electrocardiographic indices is similar in a healthy population.