Multisite near infrared spectroscopy (NIRS) monitoring during pediatric cardiopulmonary bypass (CPB) has not been extensively validated. Although it might be rational to explore regional tissue saturation at different body sites (namely brain, kidney, upper body, lower body), conflicting results are currently provided by experience in children. The aim of our study was to evaluate absolute values of multisite NIRS saturation during CPB in a cohort of infants undergoing pediatric cardiac surgery to describe average differences between cerebral, renal, upper body (arm), and lower body (thigh) regional saturation. Furthermore, the correlation between cerebral NIRS and cardiac index (CI) at CPB weaning was evaluated. Twenty-five infants were enrolled: their median weight, age, and body surface area were 3.9 (3.3-6) kg, 111 (47-203) days, and 0.24 (0.22-0.33) m(2) , respectively. Median Aristotle score was 8 (6-10), and vasoactive inotropic score at CPB weaning was 16 (14-25). A total of 17 430 data points were recorded by each sensor: two-way ANOVA showed that time (P < 0.0001) and site (P = 0.0001) significantly affected variations of NIRS values: however, if cerebral NIRS values are excluded, sensor site is no more significant (P = 0.184 in the no circulatory arrest [noCA] group and P = 0.42 in the circulatory arrest [CA] group). Analysis of NIRS saturation changes over time showed that, at all sites, average NIRS values increased after CPB start, even if the increase of cerebral saturation was less intense than other sites (P < 0.0001). Detailed analysis of interaction between site of NIRS measurement and time point showed that cerebral NIRS (ranging from 65 to 75%) was always significantly lower than that of other channels (P < 0.0001) that tended to be in the range of oversaturation (80-90%), especially during the CPB phase. Average cerebral NIRS values of patients who did not undergo circulatory arrest (CA) during CPB, 10 min after CPB weaning, were associated with average CI values with a significant correlation (r = 0.7, P = 0.003). In conclusion, during CPB, cerebral NIRS values are expected to remain constantly lower than somatic sensors, which instead tend to show similar elevated saturations, regardless of their position. Based on these results, positioning of noncerebral NIRS sensors during CPB without CA may be questioned.