The aim of this study is to investigate the relationship between invasive brain tissue oxygen pressure (PbrO(2)) and noninvasive regional transcranial oxygen saturation (rSO(2)) in 22 stable patients with severe traumatic brain injury (TBI) during a 16 h period. This was a prospective, observational study carried out in the Neurocritical Care Unit of a level 1 trauma center in a teaching hospital. A total of 41,809 paired records for neuromonitoring variables were analyzed and compared. A direct and independent correlation between rSO(2) and PbrO(2) was confirmed through adjusted [beta coefficient and (95% confidence interval, CI) = 0.36 (0.35-0.37)] and logistic [PbrO(2) >or=15 mmHg, as a dependent variable; adjusted odds ratio (AOR) and (95% CI) = 1.11 (1.10-1.12)] regression analyses. A receiver-operating characteristic (ROC) curve demonstrated that rSO(2) had low accuracy for detecting moderate (PbrO(2) <or=15 mmHg) intracerebral hypoxia [area under curve (AUC) = 0.62], with the likelihood ratio for a positive test (LR+) = 1.2 for an optimal cutoff of rSO(2) <or=70%. In contrast, the ROC analysis showed that rSO(2) was moderately accurate for detecting severe (PbrO(2) <or=12 mmHg) intracerebral hypoxemia (AUC = 0.82; LR+ = 5.3) for an optimal cutoff of rSO(2) <or=60%. In patients with severe TBI, PbrO(2) and rSO(2) were directly and significantly related. Severe intracerebral hypoxia was better detected by rSO(2) than was moderate intracerebral hypoxia. However, the diagnostic accuracy of rSO(2) was limited, and this measure should not be considered a substitute for routine PbrO(2) monitoring.