Baseline cerebral regional saturation (rSO₂) measured using the INVOS 5100C (Medtronic, MN, USA) varies widely among patients with cardiac and/or renal diseases. To identify significant correlates of baseline rSO₂ and to investigate intraoperative rSO₂ changes, we conducted a retrospective study in 494 patients undergoing on-pump cardiovascular surgery. Correlations between preoperative blood laboratory test variables and baseline rSO₂ before anesthesia were examined. Intraoperative rSO₂ changes were analyzed. Of all the variables examined, log-transformed B-type natriuretic peptide (BNP) most significantly and negatively correlated with baseline rSO₂ (r = − 0.652, p < 0.0001). Intraoperatively, rSO₂ showed the lowest value during cardiopulmonary bypass (CPB) (median rSO₂: 56.2% during CPB vs. 63.9% at baseline, p < 0.0001). Although rSO₂ during CPB correlated positively with hemoglobin concentration and oxygen delivery during CPB (r = 0.192, p < 0.0001; and r = 0.172, p = 0.0001, respectively), it correlated much more closely with baseline rSO₂ (r = − 0.589, p < 0.0001). Thus, patients showing low baseline rSO₂ primarily associated with preoperatively high BNP continued to show low rSO₂ even during CPB independent of hemodynamics artificially controlled by CPB. Our findings suggest that low baseline rSO₂ in patients with high BNP due to cardiac and/or renal diseases is more likely to result from tissue edema causing alterations in optical pathlength and thus in calculated rSO₂ values, not readily modifiable with CPB, rather than actual cerebral hemodynamic alterations readily modifiable with CPB. These may partly explain why the INVOS oximeter is a trend monitor requiring baseline measures.