Jugular bulb oxygen monitoring can be used to estimate the adequacy of cerebral blood flow to support cerebral metabolism after severe head injury. In the present study, the authors studied the cerebral arteriovenous oxygen difference (AVDO[2]) before and after treatment in 32 head-injured patients (Glasgow Coma Scale scores < or = 8) to examine the relationships among AVDO and cerebral perfusion pressure (CPP), delayed cerebral infarction, and outcome. Fifteen patients (Group A) underwent craniotomy for hematoma evacuation and 17 (Group B) received mannitol for sustained intracranial hypertension (intracranial pressure > 20 mm Hg, > 10 minutes). Radiographic evidence of delayed cerebral infarction was observed in 14 patients. Overall, 17 patients died or were severely disabled. Cerebral AVDO(2) was elevated before craniotomy or mannitol administration; the mean AVDO(2) for all patients before treatment was 8.6 +/- 1.8 vol%. Following craniotomy or mannitol administration, the AVDO(2) decreased in 27 patients and increased in five patients (mean AVDO(2) 6.2 +/- 2.1 vol% in all patients; 6 +/- 1.9 vol% in Group A; and 6.4 +/- 2.4 vol% in Group B). The mean CPP was 75 +/- 9.8 mm Hg and no relationship with AVDO(2) was demonstrated. Before treatment, the AVDO(2) was not associated with delayed cerebral infarction or outcome. By contrast, a limited improvement in elevated AVDO(2) after craniotomy or mannitol administration was significantly associated with delayed cerebral infarction (Group A: p < 0.001; Group B: p < 0.01). Similarly, a limited improvement in elevated AVDO(2) after treatment was significantly associated with an unfavorable outcome (Group A: p < 0.01; Group B: p < 0.001). In conclusion, these findings strongly indicate that, despite adequate cerebral perfusion, limited improvement in elevated cerebral AVDO(2) after treatment consisting of either craniotomy or mannitol administration may be used to help predict delayed cerebral infarction and poor outcome after traumatic brain injury.