Selective antegrade cerebral perfusion (SACP) maintains cerebral perfusion during aortic arch surgery and is mainly achieved by cannulating unilaterally (uSACP) or bilaterally (bSACP). The choice of applying uSACP or bSACP is often made according to institutional practice, whereas neurological complications may be reduced if the approaches instead were applied according to the patient’s individual cerebral collateral circulation. However, no method for predicting intraoperative cerebral collateral capacity is currently available. This study evaluates a computational method for predicting the cerebral perfusion pressure (CPP) in patients undergoing aortic arch surgery with SACP. Twenty-two patients underwent elective aortic arch surgery with bSACP, including a five-minute intervention of right-sided uSACP. Patients preoperatively underwent computed tomography angiography and 4D flow magnetic resonance imaging to assess patient-specific arterial anatomy and blood flows. Computational fluid dynamics modelling estimated the vascular resistance of the vascular beds distal of the segmented arterial anatomy. Based on these resistances, a subsequent model predicted CPP during uSACP and bSACP given intraoperatively measured input. There was no difference between predicted and measured left side CPP for uSACP (-3.0 ± 11.2 mmHg, p=0.25; r=0.75, p<0.001) or bSACP (-0.4 ± 5.7 mmHg, p=0.78; r=0.94, p<0.001). There was good group level agreement (90% CI, uSACP: [-7.5, 1.4]; bSACP: [-2.6, 1.9]) whereas limits of agreement showed substantial spread (uSACP: [-24.9, 18.9], bSACP: [-11.5, 10.8]). The new method predicted CPP on group level, while improved individual prediction accuracy may be achievable through refined arterial segmentation and more accurate estimation of intraoperative vascular resistances.