Ischaemic brain lesions still have a high prevalence in fatally head injured patients and are the single most important cause of secondary brain damage. The present study was undertaken to explore the acute phase of severely head injured patients in order to detect early ischaemia using Robertson's approach of estimating cerebral blood flow (CBF) from calculated arterio-jugular differences of oxygen (AVDO2), lactates (AVDL), and the lactate-oxygen index (LOI). Twenty-eight cases with severe head injury were included (Glasgow Coma Scale Score below or equal to 8). All patients but one had a non-missile head injury. All the patients had a diffuse brain injury according to the admission CT scan. ICP measured at the time of admission was below 20 mmHg in 17 cases (61%). All patients were evaluated with the ischaemia score (IS) devised in our center to evaluate risk factors for developing ischaemia. Mean time from injury to the first AVDO2/AVDL study was 23.9 +/- 9.9 hours. According to Robertson's criteria, 13 patients (46%) had a calculated LOI (-AVDL/AVDO2) value above or equal to 0.08 and therefore an ischaemia/infarction pattern in the first 24 hours after the accident. Of the 15 patients without the ischaemia/infarction pattern, in three cases the CBF was below the metabolic demands and therefore in a situation of compensated hypoperfusion. No patient in our series had hyperaemia. Comparing different variables in ischaemic and non-ischaemic patients, only arterial haemoglobin and ischaemia score (IS) was significantly different in both groups. The ischaemia score had mean of 4.3 +/- 1.7 in the ischaemic group and 2.7 +/- 1.4 in non-ischaemic patients (p = 0.01). It is concluded that ischaemia is highly prevalent in the early period after severe head injury. Factors potentially responsible of early ischaemia are discussed.