Brain death is defined as permanent loss of the brain functions. The evaluation of it has many meanings, such as the relief of organ transplantation stress and family burden. However, it is hard to be judged precisely. The standard clinical tests are expensive, time consuming and even dangerous, and some auxiliary methods have limitations. Functional near infrared spectroscopy (fNIRS), monitoring cerebral hemodynamic responses noninvasively, evaluate brain death in some papers published, but there is no discussion about which experimental mode can monitor brain death patient more sensitively. Here, we attempt to use our fNIRS to evaluate brain death and find which experimental mode is effective. In order to discuss the problem, we detected eleven brain death patients and twenty normal patients under natural state. They were provided different fraction of inspiration O2 (FIO2) in different phase. We found that the ratio of ∆[HbO2] (the concentration changes in oxyhemoglobin) to ∆[Hb] (the concentration changes in deoxyhemoglobin) in brain death patients is significantly higher than normal patients in FIO2 experiment. Combined with the data analysis result, restore oxygen change process and low-high-low paradigm is more sensitively.