A consecutive series of 34 severe head-injured patients (DAI) were studied prospectively. Patients were categorized according to a new, simple classification system comprised of four lesion types according to the compression or obliteration of the ventricles or cisterns. Five patients belonged to type II and 19 patients to type IV. Each type was further subdivided into two GCS score ranges (5 to 8 and below 5). The distribution of the posttraumatic infarction was mainly in the frontal and temporal lobes (60% of all cases). Our data demonstrated that the ICP was significantly lower at a 30 ° head elevation than at 0 ° (18.6 ± 7.21 mmHg vs 23.0 ± 10.60 mmHg. t = 4.22 P < 0.001), but head position did not statistically affect CPP (69.4 ± 19.86 mmHg vs 68.2 ± 19.87 mmHg. t = -0.54, P < 0.59). The effect of intensive therapy on ICP, CPP and AVDO₂ was studied in all cases, employing steroids and diuretics in a modified intensive care scale. In cases where barbiturates were employed, there were statistically significant changes in ICP and AVDO2 (P < 0.001), but CPP was not affected (P < 0.59). Surviving patients were analyzed by using the GOS and the neurological grading score (NGS, Nihon University) of the persistent vegetative state. Our data suggests that head elevation of 30 ° and barbiturate therapy are more effective on ICP and AVDO₂, and NGS more exact than GOS in vegetative patients.