Cerebral blood flow was recorded in 39 patients undergoing cardiac surgery by intraarterial injection of xenon 133. There were three subgroups of patients: 10 patients had a 20 micron arterial filter (Johnson) and 11 a 40 micron filter (Pall), and 18 had no arterial filtration. All patients had a 40 micron (Pall) filter in the coronary suction line. Significant changes in cerebral blood flow occurred during extracorporeal circulation (p < 0.0001). For all patients cerebral blood flow increased from a resting prebypass level of 30 to 46 and 57 ml/100 g a minute during initial and stable hypothermic extracorporeal circulation respectively. Both measurements were obtained at 26°C and the recordings were made on average 12 and 55 minutes after the extracorporeal circulation was started. During rewarming cerebral blood flow increased to 64, 53, 41, and 36 ml/g a minute at 310, 330, 350, and 37°C respectively, and when measured four and 16 minutes on average after bypass it was 44 and 41 ml/100 g a minute. This general brain hyperperfusion was noticed in all patients with a high enough mean blood pressure to produce hyperaemia. Interposing 20 and 40 micron arterial filters reduced cerebral blood flow but did not prevent this hyperaemia. The cerebral autoregulation, which maintains a constant cerebral blood flow within wide limits of perfusion pressures, was not affected by arterial filtration. The lower limit of blood pressure at which a further reduction in blood pressure was followed by a reduction in cerebral blood flow was around 60mmHg in all three groups.Currently postoperative mortality is of little use as a measure of the quality of cardiac surgery, as mortality rates are generally low, despite strikingly different perfusion techniques. A new index of the quality of surgery is emerging, based on psychometric analysis of the extent to which complex mental functioning is preserved in the early postoperative period, combined with more quantitative methods, such as analysis of spinal fluid intracellular markers1 -3 and measurements of cerebral blood flow during and after cardiac surgery.4-8 In our earlier studies of cerebral blood flow during cardiac surgery we found a consistent hyperperfusion of the brain during cardiopulmonary bypass6 -8 and a diffusely reduced regional cerebral blood flow after surgery,5 the most reasonable explanation being a