Objective To explore the value of intraoperative regional cerebral oxygen saturation (rSO2) monitoring in predicting postoperative delirium (POD) in infants undergoing robot assisted laparoscopic pyeloplasty, and to establish a prediction model. Methods Thirty pediatric patients, aged <4 years and American Society of Anesthesiologists (ASA) grade Ⅰ or Ⅱ, underwent robot assisted laparoscopic pyeloplasty. Monitor and record the heart rate, mean arterial pressure (MAP), pulse oxygen saturation (SpO2), rSO2 before anesthesia induction (T0), after intubation (T1), pneumoperitoneum for 5 min (T2), pneumoperitoneum for 0.5 h (T3), pneumoperitoneum for 1 h (T4), pneumoperitoneum for 2 h (T5), pneumoperitoneum for 5 min (T6), extubation for 5 min (T7) and end expiratory carbon dioxide partial pressure (PETCO2) at T1‒T6. The operation time, anesthesia time, extubation time, fentanyl dosage, liquid dosage, blood loss and urine volume during the operation were recorded. The basic value, minimum value (rSO2min) and average value (rSO2mean) of rSO2 were recorded, and the maximum percentage drop of rSO2 (rSO2%max) were calculated. The occurrence of POD in pediatric patients was evaluated within 20 min after extubation according to the Cornell Assessment of Pediatric Delirium (CAPD) scale, and they were divided into delirium group and non delirium group according to the CAPD score. Logistic regression was used to analyze the independent risk factors of delirium, receiver operating characteristic curve (ROC) was used to evaluate the predictive value of rSO2 for POD, and a monogram model was established. Results Compared with non delirium group, heart rate, MAP, SpO2 at T0‒T7 and PETCO2 at T1‒T6 in delirium group had no significant difference (P>0.05), while rSO2 at T1, T2 and T4‒T7 decreased (P<0.05). Compared with non delirium group, there was no significant difference in the basic value of rSO2 in delirium group (P>0.05), but rSO2min and rSO2mean decreased, and rSO2%max increased (P<0.05). There was no significant difference in other indicators between the two groups (P>0.05). Binary logistic regression analysis showed that rSO2%max was the influencing factor of POD (P<0.05). ROC curve analysis showed that when the critical value of rSO2%max was 3.18%, the yoden index was the largest, the sensitivity and specificity were 77.3% and 86.7% respectively, and the area under curve (AUC) of POD was 0.840. The C‑index of the monogram model was verified to be 0.735, and the trend of the calibrated prediction model was basically consistent with the ideal curve. Conclusions rSO2%max is an independent risk factor for POD during postoperative recovery in infants undergoing robot assisted laparoscopic pyeloplasty. rSO2%max>3.18% can be used as an index to predict the occurrence of POD. The monogram prediction model established by rSO2%max can predict the occurrence of POD and provide help for the prevention and diagnosis of POD.