Introduction Hyperglycemia is a marker of poor prognosis in severe brain injuries. There is currently little data regarding the effects of intensive insulin therapy (IIT) on neurological recovery. Methods A sub-group analysis of the randomized-controlled CGAO-REA study (NCT01002482) in surgical intensive care units (ICU) of two university hospitals. Patients with severe brain injury, with an expected ICU length of stay ≥48 hours were included. Patients were randomized between a conventional glucose management group (blood glucose target between 5.5 and 9 mmol.L−1) and an IIT group (blood glucose target between 4.4 and 6 mmol.L−1). The primary outcome was the day-90 neurological outcome evaluated with the Glasgow outcome scale. Results A total of 188 patients were included in this analysis. In total 98 (52%) patients were randomized in the control group and 90 (48%) in the IIT group. The mean Glasgow coma score at baseline was 7 (±4). Patients in the IIT group received more insulin (130 (68 to 251) IU versus 74 (13 to 165) IU in the control group, P = 0.01), had a significantly lower morning blood glucose level (5.9 (5.1 to 6.7) mmol.L−1 versus 6.5 (5.6 to 7.2) mmol.L−1, P <0.001) in the first 5 days after ICU admission. The IIT group experienced more episodes of hypoglycemia (P <0.0001). In the IIT group 24 (26.6%) patients had a favorable neurological outcome (good recovery or moderate disability) compared to 31 (31.6%) in the control group (P = 0.4). There were no differences in day-28 mortality. The occurrence of hypoglycemia did not influence the outcome. Conclusions In this sub-group analysis of a large multicenter randomized trial, IIT did not appear to alter the day-90 neurological outcome or ICU morbidity in severe brain injured patients or ICU morbidity.