Abstract OBJECTIVES The somatic-cerebral regional oxygen saturation (rSO2) gradient has the potential to reveal a low cardiac output state at an early stage, when systemic vascular resistance increases in order to maintain cerebral perfusion, and results in a drop in somatic rSO2. We hypothesized that during neonatal cardiopulmonary bypass (CPB), the somatic-cerebral rSO2 gradient can also signal poor perfusion, as assessed by an increase in lactate concentration. METHODS In a retrospective review of neonatal bypass records, we investigated the association between the somatic-cerebral rSO2 gradient and serial lactate measurements. The somatic-cerebral rSO2 gradient, i.e. the area between the somatic and cerebral rSO2, was calculated for each CPB time interval between 2 lactate measurements, using the trapezoidal method. The association between the somatic-cerebral rSO2 gradient and lactate variation was explored using a mixed model for repeated measurements adjusted for the CPB temperature across intervals. RESULTS A total of 94 CPB intervals were analysed in 24 neonates, median age 6 days, median weight 3.1 kg. The median lactate variation per interval was +0.2 mmol l−1 (range −1.5 to +2.7), the median somatic-cerebral gradient was 0 (range −1229% min to +2049% min) and the median CPB temperature was 29.5°C (IQR 28.1–33.4). There was an independent association between the lactate increase and the somatic-cerebral rSO2 gradient (regression coefficient = −0.0006 ± 0.0002, P = 0.009). CONCLUSIONS A decrease in the somatic-cerebral rSO2 gradient was associated with an increase in lactate concentration, suggesting that the somatic-cerebral rSO2 gradient is a useful monitoring tool for the adequacy of perfusion during neonatal CPB.