OBJECTIVE To minimize the neurological complications following cardiovascular surgery, it is essential to prevent an occurrence of cerebrovascular embolism and to detect and solve cerebral malperfusion without delay in the operating theater. Although we have introduced near-infrared spectroscopy (NIRS) monitoring for the purpose of detecting cerebral malperfusion, no criterion has been available. We searched for this criterion by examining the relationship of sustained drop in the regional oxygen saturation (rSO2) of the frontal lobes to the occurrence of neurological events. METHODS The 59 consecutive patients undergoing aortic surgery with selective cerebral perfusion (SCP) were examined. The rSO2 was monitored throughout the surgery and the durations of drops in rSO2 to below 55% and those below 60% were determined for each patient. The durations of rSO2 drop and other surgery-related parameters were compared between the patients in whom neurological events occurred and those without such events. RESULTS A total of 16 cases (27.1%) presented with neurological events. Newly developed cerebral infarction was documented in 6 of these 16 cases. Operation time and the durations for which rSO2 dropped were significantly longer for the 16 patients with neurological events than for the 43 patients without events (Op time: 546.8 versus 448.1 min, P=0.0064; rSO2 below 60%: 141.2 versus 49.8 min, P=0.0032; rSO2 below 55%: 66.6 versus 10.6 min, P=0.0011), while there was no significant difference in age, bypass time, aortic clamping time, SCP time, and circulatory arrest time between the two groups. In the 3 patients with infarcts suggestive to hypoperfusion, sustained decrease in rSO2 was observed, while it was not significant in the remaining 3 patients with infarcts suggestive to embolism. Among the 53 patients without infarction, transient neurological events occurred more frequently in patients with sustained drop in rSO2 below 55% for over 5 min (44.4% versus 5.7%, P=0.0014). CONCLUSIONS A sustained drop in rSO2 during aortic surgery is closely related to the occurrence of neurological events following surgery. We recommend that recovery of drop in rSO2 below 55% should be addressed without delay. However, use of NIRS is limited for detecting embolic events or hypoperfusion in the basilar region.