BACKGROUND: Optimal glycaemic targets for patients with severe traumatic brain injury remain unclear. The primary objective of this microdialysis study was to compare cerebral metabolism with strict versus conventional glycaemic control. METHODS: We performed a prospective single-centre randomised controlled within-subject crossover study of 20 adult patients admitted to an academic neurointensive care unit with severe traumatic brain injury. Patients underwent randomised, consecutive 24-h periods of strict (4-7 mmol/L; 72-126 mg/dl) and conventional (<10 mmol/L; 180 mg/dl) glycaemic control with microdialysis measurements performed hourly. The first 12 h of each study period was designated as a 'washout' period, with the subsequent 12 h being the period of interest. RESULTS: Cerebral glucose was lower during strict glycaemia than with conventional control (mean 1.05 [95% CI 0.58-1.51] mmol/L versus 1.28 [0.81-1.74] mmol/L; P = 0.03), as was lactate (3.07 [2.44-3.70] versus 3.56 [2.81-4.30]; P < 0.001). There were no significant differences in pyruvate or the lactate/pyruvate ratio between treatment phases. Strict glycaemia increased the frequency of low cerebral glucose (< 0.8 mmol/L; OR 1.91 [95% CI 1.01-3.65]; P < 0.05); however, there were no differences in the frequency of critically low glucose (< 0.2 mmol/L) or critically elevated lactate/pyruvate ratio between phases. CONCLUSIONS: Compared with conventional glycaemic targets, strict blood glucose control was associated with lower mean levels of cerebral glucose and an increased frequency of abnormally low glucose levels. These data support conventional glycaemic targets following traumatic brain injury. TRIAL REGISTRATION: ISRCTN, ISRCTN19146279 . Retrospectively registered on 2 May 2014.