Intraparenchymal haemorrhages occur in a proportion of severe head-injured patients but the role of surgery in their treatment is unclear. This international multi-centre patient-randomised parallel-group trial compared early surgery (haematoma evacuation within 12 hours of randomisation) with initial conservative treatment (later evacuation allowed if deemed necessary). Patients were randomised using an independent randomisation service within 48 hours of head injury. Patients were eligible if they had no more than two intraparenchymal haemorrhages of 10 ml or more and did not have an extradural or subdural haematoma that required surgery. The primary outcome measure was the traditional dichotomous split of the Glasgow Outcome Scale obtained by postal questionnaires sent directly to patients at 6 months. The trial was halted early by the UK funding agency (NIHR HTA) for failure to recruit sufficient patients from the UK. (Trial registration: ISRCTN19321911.) A total of 170 patients were randomised from 31 of 59 registered centres worldwide. Of 82 patients randomised to early surgery with complete follow up 30 (37%) had an unfavourable outcome. Of 85 patients randomised to initial conservative treatment with complete follow up 40 (47%) had an unfavourable outcome (odds ratio 0.65 [95% CI 0.35, 1.21], p=0.17); absolute benefit 10.5% (-4.4% to 25.3%). There were significantly more deaths in the first six months in the initial conservative treatment group (33% v 15%; p = 0.006). The 10.5% absolute benefit with early surgery was consistent with the initial power calculation. However, with the low sample size resulting from the premature termination, we cannot exclude the possibility that this could be a chance finding. A further trial is required urgently to assess whether this encouraging signal can be confirmed.