BACKGROUND AND OBJECTIVE: In-vitro performance of the PediaSat system for continuous monitoring of central venous oxygen saturation by spectrophotometry has been evaluated. METHODS: PediaSat continuous fibre-optic oximetry catheters were inserted in a black testing chamber, connected with an extracorporeal circuit and filled with human whole blood. Oxygen inflow into the cardiopulmonary bypass system was varied, and the testing chamber was perfused with blood flow of 1000 ml min. Oxygen saturation values measured by PediaSat (SPediaSatO2) were compared with cooximetry (SCO-OXO2) values from simultaneously taken blood samples by Bland-Altman and simple regression analyses. RESULTS: Fifty data pairs were obtained. SPediaSatO2 and SCO-OXO2 values ranged between 28-98 and 24.9-99.5%, respectively. Correlation between SPediSatO2 and SCO-OXO2 was high with an r value equal to 0.96 (P < 0.0001). Overall, SPediaSatO2 only slightly overestimated SCO-OXO2 (mean bias +2.9%), and limits of agreement (+/-2 SD of bias) were acceptable (-6.8/+12.6%). Sensitivity and specificity of the first differences of SPediaSatO2 and SCO-OXO2 were 1.0 and 0.92, respectively. Subgroup analysis of SCO-OXO2 values below 70% resulted in an overestimation by SPediaSatO2, with a mean bias of +5.2% and limits of agreement of -4.7 and +15.1%. CONCLUSION: The current version of the PediaSat system does not reliably reflect SCO-OXO2 values below 70%, but it seems to be a useful tool providing an accurate trend of continuous central venous oxygen saturation.