Background: To date there has been no reliable “real time” monitoring available to determine the adequacy of oxygen delivery and cerebral perfusion during cardiac arrest. Although, we have previously demonstrated that the mean regional cerebral oxygen saturation (rSO2) measured by cerebral oximetry can be used to determine return of spontaneous circulation in cardiac arrest, the rSO2 level that corresponds with ROSC in the 3 main different subtypes of cardiac arrest (VF/VT, PEA, asystole has not been examined. Objectives: To investigate the level of rSO2 that corresponds with ROSC in 1) VF/VT, 2) PEA and 3) asystole subtypes of cardiac arrest using a commercially available cerebral oximeter (Equanox, Nonin, USA) Methods: Cerebral oximetry was applied to 34 in-hospital cardiac arrest patients and continuous data was collected until either ROSC was achieved or CPR was terminated. Results: Overall 20 patients achieved ROSC, whereas 14 did not. There was a significant overall difference in mean±SD rSO2% in patients who achieved ROSC (49.7 ± 11.3% vs. 33.9 ± 15.0% p=<0.001) compared to those who did not. Examination of the cardiac arrest subtypes revealed that while this difference was observed in the PEA (n=13) 54.6 ± 8.7 vs. 36.8 ± 20.4 (p=<0.05) and asystole (n=16) 46.6 ± 11.4 vs. 27.4 ± 10.8 (p=<0.05) subgroups, there was no difference in rSO2 in patients with and without ROSC in the VF/VT subgroup (n=5) 41.6 ± 18.5 vs. 45.0 ± 11.6 (p=0.8). Conclusions : Although cerebral oximetry may be used as a real-time non-invasive marker of ROSC during cardiac arrest, the main difference in rSO2 between patients achieving ROSC and those without ROSC applies to the asystole and PEA sub groups of cardiac arrest, rather than VF/VT. This likely reflects the different dominant physiological disorders between shockable and non-shockable causes of cardiac arrest. Whereas recovery from PEA and asystole is dependent on the adequacy of oxygen delivery, recovery from VF/VT predominantly relates to the timing of defibrillation.