BACKGROUND Transcranial Doppler sonography is increasingly used to monitor changes in cerebral perfusion intraoperatively. However, little information is available about the validity of velocity measurements as an index of cerebral blood flow (CBF). The purpose of this study was to compare invasive and Doppler-derived measurements of cerebral hemodynamic variables during coronary artery bypass graft surgery. METHODS In 15 male patients, measurements of CBF and middle cerebral artery flow velocity (VMCA) were performed before and after induction of fentanyl-midazolam anesthesia, during hypothermic cardiopulmonary bypass (CPB), and at the end of the surgical procedure. Transcranial Doppler sonography recordings of systolic, diastolic, and mean VMCA, and derived parameters such as pulsatility (PI) and resistance (RI) indexes were recorded from the proximal segment of the right middle cerebral artery. CBF was measured by the Kety-Schmidt inert gas saturation method with argon as a tracer. To facilitate comparisons of CBF and VMCA measurements, changes between consecutive measurements were expressed as percentage values. Calculations of cerebral perfusion pressure and cerebral vascular resistance (CVR) were based on jugular bulb pressure. The cerebral metabolic rate for oxygen was calculated from CBF and the arterial-cerebral venous oxygen content difference. RESULTS Changes in mean VMCA paralleled changes in mean CBF except for hemodynamic changes associated with hypothermic CPB. At this stage of surgery, mean VMCA increased while actual CBF decreased. Separate analysis of the periods before and after CPB revealed a poor association between percentage changes in CBF and VMCA (r = 0.26, P = 0.36; r = 0.51, P = 0.06, respectively). Mean values of CVR, PI, and RI showed consistent changes after induction of anesthesia. After termination of CPB, mean CVR significantly decreased, whereas mean PI and RI remained virtually unchanged. Neither before nor after CPB was a clinically useful correlation found between percentage changes in PI, RI, and CVR (PI r = 0.28, P = 0.34; r = -0.47, P = 0.09, respectively; RI r = 0.16, P = 0.59; r = -0.53, P = 0.06, respectively). CONCLUSIONS Hypothermic CPB seems to alter the relation between global CBF and flow velocity in basal cerebral arteries. Inconsistency in directional changes in CBF and VMCA at this stage of surgery might be attributable to changes in middle cerebral artery diameter, red blood cell velocity spectra, and regional flow distribution. Although changes in mean VMCA before and after CPB appear to parallel changes in mean CBF, individual responses of VMCA cannot reliably predict percentage changes in CBF. Furthermore, Doppler sonographic PI and RI cannot provide an approximation of changes in CVR during cardiac surgery.