Background. Near-infrared spectroscopy (NIRS) has been highlighted as a potentially useful non-invasive clinical tool for continuously monitoring cerebral hemodynamics in critically ill patients with sepsis (Steiner et al, 2009). This is based on the presupposition that changes in cerebral oxygen saturation (ScO2) accurately reflect changes in cerebral blood flow (CBF), but whether this is the case in critically ill patients with sepsis remains to be clarified. In the present study, we therefore compared NIRS- to transcranial Doppler ultrasound (TCD)-derived estimates of vasopressor-associated changes in CBF. Methods. Eight mechanically ventilated, critically ill patients (mean age 62, SD 11; 2 females) diagnosed with severe sepsis were included following ethical approval and informed consent from the next of kin. In a step-wise manner, the mean arterial blood pressure (MAP) was increased by approximately 30 mmHg over 4-5 steps, remaining 5 minutes at each step, by increasing the norepinephrine infusion rate. The associated changes in CBF were assessed by simultaneous ipsilateral NIRS (ScO2) and TCD (middle cerebral artery blood flow velocity, MCAv) measurements. Data are reported as median with corresponding interquartile range (IQR). Results. A total of fifteen simultaneous NIRS- and TCD-derived assessments of vasopressors-associated changes in CBF were obtained in the patients. The norepinephrine infusion rate was raised from 0 median (IQR, 0-0 μg/kg/min) to 0.20 μg/kg/min median (IQR, 0.15-0.30 μg/kg/min), which increased MAP from 74 (71-90) to 100 (93-115) mmHg (p<0.05). This was associated with an increase in MCAv of 14% (2-22) (p<0.05), whereas no changes were observed in ScO2; 1% ([-4]-3) (NS). The Bland-Altman plot (Bland & Altman, 1986), used to compare the two methods, showed a relative bias of 14% with limits of agreement -16 to 44 % change in CBF (Figure). Conclusion. There was poor agreement between NIRS- and TCD-derived estimates of norepinephrine-induced changes in CBF. This is likely related to contamination of the signal by extracranial tissues, particularly blood flow through the scalp vessels. Our findings thus question the applicability of NIRS for monitoring cerebral hemodynamics, at least in critically ill patients with sepsis.