IMPORTANCE: Ischemia is an important pathophysiological mechanism after traumatic brain injury (TBI), but its incidence and spatiotemporal patterns are poorly characterized. OBJECTIVE: To comprehensively characterize the spatiotemporal changes in cerebral physiology after TBI. DESIGN, SETTING, AND PARTICIPANTS: This single-center cohort study uses 15oxygen positron emission tomography data obtained in a neurosciences critical care unit from February 1998 through July 2014 and analyzed from April 2018 through August 2019. Patients with TBI requiring intracranial pressure monitoring and control participants were recruited. EXPOSURES: Cerebral blood flow (CBF), cerebral blood volume (CBV), cerebral oxygen metabolism (CMRO2), and oxygen extraction fraction. MAIN OUTCOMES AND MEASURES: Ratios (CBF/CMRO2 and CBF/CBV) were calculated. Ischemic brain volume was compared with jugular venous saturation and brain tissue oximetry. RESULTS: A total of 68 patients with TBI and 27 control participants were recruited. Results from 1 patient with TBI and 7 health volunteers were excluded. Sixty-eight patients with TBI (13 female [19%]; median [interquartile range (IQR)] age, 29 [22-47] years) underwent 90 studies at early (day 1 [n = 17]), intermediate (days 2-5 [n = 54]), and late points (days 6-10 [n = 19]) and were compared with 20 control participants (5 female [25%]; median [IQR] age, 43 [31-47] years). The global CBF and CMRO2 findings for patients with TBI were less than the ranges for control participants at all stages (median [IQR]: CBF, 26 [22-30] mL/100 mL/min vs 38 [29-49] mL/100 mL/min; P < .001; CMRO2, 62 [55-71] μmol/100 mL/min vs 131 [101-167] μmol/100 mL/min; P < .001). Early CBF reductions showed a trend of high oxygen extraction fraction (suggesting classical ischemia), but this was inconsistent at later phases. Ischemic brain volume was elevated even in the absence of intracranial hypertension and highest at less than 24 hours after TBI (median [IQR], 36 [10-82] mL), but many patients showed later increases (median [IQR] 6-10 days after TBI, 24 [4-42] mL; across all points: patients, 10 [5-39] mL vs control participants, 1 [0-3] mL; P < 001). Ischemic brain volume was a poor indicator of jugular venous saturation and brain tissue oximetry. Patients' CBF/CMRO2 ratio was higher than controls (median [IQR], 0.42 [0.35-0.49] vs 0.3 [0.28-0.33]; P < .001) and their CBF/CBV ratio lower (median [IQR], 7.1 [6.4-7.9] vs 12.3 [11.0-14.0]; P < .001), suggesting abnormal flow-metabolism coupling and vascular reactivity. Patients' CBV was higher than controls (median [IQR], 3.7 [3.4-4.1] mL/100 mL vs 3.0 [2.7-3.6] mL/100 mL; P < .001); although values were lower in patients with intracranial hypertension, these were still greater than controls (median [IQR], 3.7 [3.2-4.0] vs 3.0 [2.7-3.6] mL/100 mL; P = .002), despite more profound reductions in partial pressure of carbon dioxide (median [IQR], 4.3 [4.1-4.6] kPa vs 4.7 [4.3-4.9] kPa; P = .001). CONCLUSIONS AND RELEVANCE: Ischemia is common early, detectable up to 10 days after TBI, possible without intracranial hypertension, and inconsistently detected by jugular or brain tissue oximetry. There is substantial between-patient and within-patient pathophysiological heterogeneity; ischemia and hyperemia commonly coexist, possibly reflecting abnormalities in flow-metabolism coupling. Increased CBV may contribute to intracranial hypertension but can coexist with abnormal CBF/CBV ratios. These results emphasize the need to consider cerebrovascular pathophysiological complexity when managing patients with TBI.