Aims: Evaluation of the neonatal QT interval is important to diagnose arrhythmia syndromes and evaluate side effects of drugs. We aimed at describing the natural history of the QT interval duration during the first 4 weeks of life and to provide reference values from a large general population sample. Methods and results: The Copenhagen Baby Heart Study is a prospective general population study that offered cardiac evaluation of newborns. Eight-lead electrocardiograms were obtained and analysed with a computerized algorithm with manual validation. We included 14 164 newborns (52% boys), aged 0-28 days, with normal echocardiograms. The median values (ms, 2-98%ile) for the corrected intervals QTc (Bazett), QTc (Hodges), QTc (Fridericia), and QTc (Framingham) were 419 (373-474), 419 (373-472), 364 (320-414), and 363 (327-405). During the 4 weeks, we observed a small decrease of QTcFramingham, and an increase of QTcHodges (both P < 0.01), while QTcBazett and QTcFridericia did not change (P > 0.05). Applying published QT interval cut-off values resulted in 5-25% of the newborns having QT prolongation. Uncorrected QT intervals decreased linearly with increasing heart rate (HR). Sex and infant size did not affect the QT interval and the gestational age (GA) only showed an effect when comparing the extreme low- vs. high GA groups (≤34 vs. ≥42 weeks, P = 0.021). Conclusion: During the 4 weeks QTcFramingham and QTcHodges showed minor changes, whereas QTcBazett and QTcFridericia were stable. The QT interval was unaffected by sex and infant size and GA only showed an effect in very premature newborns. Reference values for HR-specific uncorrected QT intervals may facilitate a more accurate diagnosis of newborns with abnormal QT intervals.