In the United Kingdom most head injury patients are first admitted to a District General Hospital (DGH); selected patients are later referred to a Neurosurgical Unit (NSU). It is known that this system leads to some avoidable deaths. In an attempt to discover whether the minimum amount of time involved in such a system of secondary referral must of necessity put some patients at risk, and whether any significant delay could be eliminated without radical change in this system, we have carried out a detailed prospective analysis of the sequence of events involved in the emergency transfer of 117 consecutive patients with suspected traumatic intracranial haematoma (TICH). We discovered that once it had been decided to transfer a patient, the actual process of transfer consumed relatively little time regardless of the distance from the DGH to the NSU. Where harmful avoidable delay did occur was at the DGH itself either from failure to institute appropriate treatment for non-cranial injuries or from failure to realise that transfer was necessary. We believe that the geographical dispersal of neurosurgical services would not improve the outlook of patients with head injury. The optimum outcome could be achieved by concentrating head injury admissions at selected DGH's each of which would be equipped both with a CT scanner linked to a neurosurgical unit and a standby ambulance for transferring head injury patients.(ABSTRACT TRUNCATED AT 250 WORDS)