The objective of this study was to evaluate the beat-to-beat variability of the QT interval during the day and night. A new algorithm was used to detect the onset of the QRS, the apex of the T wave, and end of the T in ambulatory electrocardiographic recordings. Beat-to-beat variability of QT, QaT, and QTc during the day and night was studied in the time, frequency, and chaotic domains. Participants were adults without clinical evidence of heart disease. Although the QT duration was higher (p = 0.0001) at night, the beat-to-beat variability of this interval was lower: in the time domain (decreased standard deviation, p = 0.0005), in the frequency domain (decreased low-frequency power of the spectra, p = 0.004), and the chaotic domain (tighter clustering of the points in the Poincaré plots). The high-frequency to low-frequency ratio of the power spectra of the QT (and the RR) was higher (p = 0.03) at night. Beat-to-beat QT variability in the time, frequency, and chaotic domains is decreased at night with shift of the QT modulation to higher frequencies corresponding to respiration and representing vagal preponderance. The techniques presented here and the findings in normal subjects may be useful in evaluating the risk for arrhythmic events in patients with heart disease.