Patients with severe head injury frequently have evidence of elevated intracranial pressure (ICP) and ischemic neuronal damage at autopsy. Mannitol has been used clinically to reduce ICP with varying success, and it is possible that it is more effective in some types of head injury than in others. The aim of the present study was to determine the effect of mannitol on ICP, cerebral perfusion pressure (CPP), and cerebral blood flow (CBF) in patients with severe head injury, and to discover if these effects differed in different types of injury. Measurements of CPP, ICP, and CBF were made in 55 patients with severe head injury. In general, the resting level of CBF was higher in patients with diffuse injury (mean 50.2 ml/100 gm/min) than in those with focal injury (mean 39.8 ml/100 gm/min). Mannitol consistently reduced ICP and increased CPP and CBF by 10 to 20 minutes after infusion. The lowest flows (31.8 ml/100 gm/min) were recorded from the most damaged hemispheres of patients with focal injuries and elevated ICP. The baseline levels of flow did not correlate with ICP, CPP, Glasgow Coma Scale score, or outcome. Only four of the 55 patients had a CBF of less than 20 ml/100 gm/min in either or both hemispheres. The few low CBF's in this and other studies may reflect the steady-state conditions under which measurements are made in intensive care units, and that these patients have entered a phase of reperfusion.