Regional cerebral oxygen saturation (ScO2) determined by near-infrared spectroscopy, monitoring both arterial and venous blood oxygenation of the brain, could reflect the balance between oxygen delivery and consumption. The aim of this study was to determine the predictabilities of ScO2 and estimated oxygen extraction ratio (eO2ER) with outcomes in infants with congenital heart disease (CHD). This study was a two-center, retrospective study of patients at 12 months of age or younger with CHD who underwent cardiac surgery. The primary outcome was a composite of one or more major adverse events (MAEs) after surgery: death from any cause, circulatory collapse that needed cardiopulmonary resuscitation, and requirement for extracorporeal membrane oxygenation. Based on the assumptions of arterial to venous blood ratio, eO2ER was calculated. A total of 647 cases were included in this study. MAEs occurred in 16 patients (2.5%). There were significant differences in post-bypass ScO2 [46.61 (40.90, 52.05) vs. 58.52 (51.52, 66.08), p < 0.001] and post-bypass eO2ER [0.66 (0.60, 0.78) vs. 0.52 (0.43, 0.61), p < 0.001] between patients with MAEs and patients without MAEs. Area under the receiver operating curve (AUROC) of post-bypass ScO2 was 0.818 (95% confidence interval: 0.747–0.889), AUROC of post-bypass eO2ER was 0.783 (0.697–0.870) and AUROC of post-bypass maximum serum lactate level was 0.635 (0.525–0.746). Both ScO2 and eO2ER, especially after weaning off bypass, are acceptable predictive markers for predicting MAEs after cardiac surgery in infants.