Severe traumatic brain injury (TBI) is one of the major causes of death in younger age groups. In Umeå, Sweden, an intracranial pressure (ICP) targeted therapy protocol, the Lund concept, has been used in treatment of severe TBI since 1994. Decompressive craniectomy is used as a protocolguided treatment step. The primary aim of the investigation was to study the effect of craniectomy on ICP changes over time in patients with severe TBI treated by an ICP-targeted protocol. In this retrospective study, all patients treated for severe TBI during 1998-2001 who fulfilled the following inclusion criteria were studied: GCS Յ 8 at intubation and sedation, first recorded cerebral perfusion pressure (CPP) of Ͼ10 mm Hg, arrival within 24 h of trauma, and need of intensive care for Ͼ72 h. Craniectomy was performed when the ICP could not be controlled by evacuation of hematomas, sedation, ventriculostomy, or low-dose pentothal infusion. Ninety-three patients met the inclusion criteria. Mean age was 37.6 years. Twenty-one patients underwent craniectomy as a treatment step. We found a significant reduction of the ICP directly after craniectomy, from 36.4 mm Hg (range, 18-80 mm Hg) to 12.6 mm Hg (range, 2-51 mm Hg). During the following 72 h, we observed an increase in ICP during the first 8-12 h after craniectomy, reaching approximately 20 mm Hg, and later levelling out at approximately 25 mm Hg. The reduction of ICP was statistically significant during the 72 h. The outcome as measured by Glasgow Outcome Scale (GOS) did not significantly differ between the craniectomized group (DC) and the non-craniectomized group (NDC). The outcome was favorable (GOS 5-4) in 71% in the craniectomized group, and in 61% in the noncraniectomized group. Craniectomy is a useful tool in achieving a significant reduction of ICP over time in TBI patients with progressive intracranial hypertension refractory to medical therapy. The procedure seems to have a satisfactory effect on the outcome, as demonstrated by a high rate of favorable outcome and low mortality in the craniectomized group, which did not significantly differ compared with the non-craniectomized group.