In a study of the cerebral consequences of cardiopulmonary bypass (CPB), we have assessed cerebral blood flow (CBF) by intra-arterial Xenon 133 clearance, utilising a Novocerebrograph 10a, before, during and immediately following CPB. All patients (n = 51), under the care of a single surgical team underwent elective coronary revascularisation using a standardised anaesthetic and surgical protocol. CBF, measured as the initial slope index (ISI), fell significantly from before (median 22.5) to on bypass (median 20) (Wilcoxon P less than 0.005) and was significantly increased in the immediate post-bypass period (median 28) compared with pre-bypass (Wilcoxon P less than 0.001). Scattergrams reveal CBF to be independent of arterial pressure (BP) but show an important relationship between arterial PaCO2 and CBF. The correlations between PaCO2 and CBF prior to bypass (r = 0.46 P less than 0.005) and post-bypass (r = 0.46 P less than 0.001) are very similar, whilst on bypass, the correlation, although remaining significant, is reduced (r = 0.31 P less than 0.02). The median values for arterial PaCO2 are low throughout the study (pre-bypass median 33 mmHg, on bypass median 28 mm Hg and post-bypass median 36 mm Hg). The maintenance of PaCO2 within the normal range of 35-45 mm Hg may minimise the risk of low CBF due to hypocarbia which could theoretically be harmful in patients known to be at risk of both short term and long term cerebral dysfunction.