Although the use of on-line monitoring of brain ti-pO2 is increasing, so far the critical level of 10 mmHg is derived from animal experiments and clinical analyses: no hard proof on outcome basis has been given until now. The authors present an outcome analysis of 35 patients with severe head injury. Inclusion criteria were: start of ti-pO2 monitoring < or = 40 h post-injury, the probe lying in CT scan normal tissue and the GOS at 6 months being available. The good outcome group (GOS 4 + 5, n = 17) showed a 17.7+/-9.1 h delay from the injury to the monitoring compared to the bad outcome group (GOS 1-3, n = 18) with (14.2+/-9.1 h) (p < 0.05). Age and initial Glasgow Coma Score were not different. In the bad outcome group there were more patients with a diffuse injury type 3 and 4. The distribution of the ti-pO2 values show in all the examined time intervals (day 0-6, 0-72 h, 0-48 h and 0-24 h) a left shift in the bad outcome group with most pronounced difference for ti-pO2 < or = 10 mmHg. For the period from 0-48 h and even more from 0-24 h post-injury, the difference between both groups was significant (p = 0.036 and p = 0.013). In the bad outcome group 35.5% of the values from 0-24 h were < 10mmHg (compared to 10.6% in the good outcome group. ti-pO2 values > or = 50 mmHg were seen more often in the bad outcome group; this occurred mainly after 48-72 h post-injury. The authors concluded that brain ti-pO2 monitoring is able to detect the occurrence of early hypoxic insults. Brain ti-pO2 monitoring is an important parameter in the multimodality monitoring system.