BACKGROUND: For critically ill adult patients with acute traumatic brain injury (TBI), we assessed the clinical and cost-effectiveness of: (a) Management in dedicated neurocritical care units versus combined neuro/general critical care units within neuroscience centres. (b) 'Early' transfer to a neuroscience centre versus 'no or late' transfer for those who present at a non-neuroscience centre. METHODS: The Risk Adjustment In Neurocritical care (RAIN) Study included prospective admissions following acute TBI to 67 UK adult critical care units during 2009-11. Data were collected on baseline case-mix, mortality, resource use, and at six months, Glasgow Outcome Scale Extended (GOSE), and quality of life (QOL) (EuroQol 5D-3L). We report incremental effectiveness, costs and cost per Quality-Adjusted Life Year (QALY) of the alternative care locations, adjusting for baseline differences with validated risk prediction models. We tested the robustness of results in sensitivity analyses. FINDINGS: Dedicated neurocritical care unit patients (N = 1324) had similar six-month mortality, higher QOL (mean gain 0.048, 95% CI -0.002 to 0.099) and increased average costs compared with those managed in combined neuro/general units (N = 1341), with a lifetime cost per QALY gained of £14,000. 'Early' transfer to a neuroscience centre (N = 584) was associated with lower mortality (odds ratio 0.52, 0.34-0.80), higher QOL for survivors (mean gain 0.13, 0.032-0.225), but positive incremental costs (£15,001, £11,123 to £18,880) compared with 'late or no transfer' (N = 263). The lifetime cost per QALY gained for 'early' transfer was £11,000. CONCLUSIONS: For critically ill adult patients with acute TBI, within neuroscience centres management in dedicated neurocritical care units versus combined neuro/general units led to improved QoL and higher costs, on average, but these differences were not statistically significant. This study finds that 'early' transfer to a neuroscience centre is associated with reduced mortality, improvement in QOL and is cost-effective.