Background: The indications for intracranial pressure (ICP) monitoring in patients with acute brain injury and the effects of ICP on patients’ outcomes are uncertain. The aims of this study were to describe current ICP monitoring practises for patients with acute brain injury at centres around the world and to assess variations in indications for ICP monitoring and interventions, and their association with long-term patient outcomes. Methods: We did a prospective, observational cohort study at 146 intensive care units (ICUs) in 42 countries. We assessed for eligibility all patients aged 18 years or older who were admitted to the ICU with either acute brain injury due to primary haemorrhagic stroke (including intracranial haemorrhage or subarachnoid haemorrhage) or traumatic brain injury. We included patients with altered levels of consciousness at ICU admission or within the first 48 h after the brain injury, as defined by the Glasgow Coma Scale (GCS) eye response score of 1 (no eye opening) and a GCS motor response score of at least 5 (not obeying commands). Patients not admitted to the ICU or with other forms of acute brain injury were excluded from the study. Between-centre differences in use of ICP monitoring were quantified by using the median odds ratio (MOR). We used the therapy intensity level (TIL) to quantify practice variations in ICP interventions. Primary endpoints were 6 month mortality and 6 month Glasgow Outcome Scale Extended (GOSE) score. A propensity score method with inverse probability of treatment weighting was used to estimate the association between use of ICP monitoring and these 6 month outcomes, independently of measured baseline covariates. This study is registered with ClinicalTrial.gov, NCT03257904. Findings: Between March 15, 2018, and April 30, 2019, 4776 patients were assessed for eligibility and 2395 patients were included in the study, including 1287 (54%) with traumatic brain injury, 587 (25%) with intracranial haemorrhage, and 521 (22%) with subarachnoid haemorrhage. The median age of patients was 55 years (IQR 39–69) and 1567 (65%) patients were male. Considerable variability was recorded in the use of ICP monitoring across centres (MOR 4·5, 95% CI 3·8–4·9 between two randomly selected centres for patients with similar covariates). 6 month mortality was lower in patients who had ICP monitoring (441/1318 [34%]) than in those who were not monitored (517/1049 [49%]; p<0·0001). ICP monitoring was associated with significantly lower 6 month mortality in patients with at least one unreactive pupil (hazard ratio [HR] 0·35, 95% CI 0·26–0·47; p<0·0001), and better neurological outcome at 6 months (odds ratio 0·38, 95% CI 0·26–0·56; p=0·0025). Median TIL was higher in patients with ICP monitoring (9 [IQR 7–12]) than in those who were not monitored (5 [3–8]; p<0·0001) and an increment of one point in TIL was associated with a reduction in mortality (HR 0·94, 95% CI 0·91–0·98; p=0·0011). Interpretation: The use of ICP monitoring and ICP management varies greatly across centres and countries. The use of ICP monitoring might be associated with a more intensive therapeutic approach and with lower 6-month mortality in more severe cases. Intracranial hypertension treatment guided by monitoring might be considered in severe cases due to the potential associated improvement in long-term clinical results. Funding: University of Milano-Bicocca and the European Society of Intensive Care Medicine.