The aim of this study was to evaluate the effects of permissive hypercapnia on Cerebral Oxygen Metabolism (COM) and brain function in patients with craniocerebral trauma surgery. Sixty patients with severe traumatic brain injury were randomly divided into three groups (n=20), hypocapnia (group L, PaCO2 25-30 mmHg), hypercapnia (group H, PaCO2 50-55 mmHg), and control (group C, PaCO2 35-45 mmHg) groups. All patients underwent retrograde jugular vein and radial artery puncturing. The inducing drugs used were midazolam 0.1 mg/kg, sufentanil 0.5 μg/kg, vecuronium 0.1 mg/kg, and propofol 1.5 mg/kg. Blood samples from an artery and vein were extracted to measure the oxygen content at three time points, during cutting of the endocranium (T1), 1 h after the cutting (T2), suturing of the endocranium (T3). We measured the vein ball oxygen content difference (Ca-vDO2) to observe COM. Glasgow scores were recorded at preoperation and postoperative 24 h and 10 d. Compared with group C, the Ca-vDO2 of group L and H decreased at T2 and T3 (P<0.05); compared with T1, Ca-vDO2 decreased at T2 and T3 for all patients (P<0.05). Glasgow scores of the three groups were not statistically different at preoperation and postoperative 24 h (P>0.05). After 10 d, compared with group L, patients with a 3-point score decreased significantly and the 3-8 point patients increased significantly in group H (P<0.05); compared with group C, the 3-point score patients decreased significantly in group H (P<0.05). Permissive hypercapnia may improve Glasgow scores and prognosis of patients of severe traumatic brain injury without affecting oxygen uptake of brain.