The effects on intracranial pressure (ICP), cerebral perfusion pressure (CPP) and EEG monitored by Cerebral Function Monitor (CFM) were compared after bolus administration of mannitol (n = 55) and thiopentone (n = 67) to control intracranial hypertension in 18 severely head injured patients. Mannitol increased CPP in 89% of occasions and thiopentone in only 54% (p < 0.001). Thiopentone caused a mean increase in CPP +0.6 kPa (+5.0 +/- 1.6 mmHg) when the minimum pre-bolus voltage of CFM was above 4 microV and a fall in CPP -0.5 kPa (-4.1 +/- 1.6 mmHg) when cortical electrical activity was already severely depressed (p < 0.0002). When pre-bolus CFM was below 4 microV mannitol was superior to thiopentone. This different effect on CPP was due to a significantly greater fall in mean arterial pressure in the thiopentone sub-group -1.6 versus -0.3 kPa (-12.4 +/- 1.5 mmHg, -2.8 +/- 1.2 mmHg; p < 0.001). Severe and unpredictable hypotension occurred, particularly in the thiopentone low CFM sub-group. This symptomatic therapy seems inadequate but a targeted treatment of intracranial hypertension could be possible only with a more sophisticated monitoring, including continuous data on cerebral blood flow and adequacy to metabolic demand.