OBJECTIVE: To evaluate the Medical Disability Society's 1988 recommendation that "every patient attending hospital after a head injury should be registered and offered an outpatient follow up appointment" by determining whether offering a routine follow up service to patients presenting to hospital with a head injury of any severity affects outcome six months later. DESIGN: A randomised controlled trial design with masked assessment of outcome. SETTING: A mixed rural and urban health district with a population of about 560000. PATIENTS: 1156 consecutive patients resident in Oxfordshire aged between 16 and 65 years presenting over 13 months to accident and emergency departments or admitted to hospital and diagnosed as having a head injury of any severity, including those with other injuries. INTERVENTIONS: Patients were registered and randomised to one of two groups. Both groups continued to receive the standard service offered by the hospitals. The early follow up group were approached at 7-10 days after injury and offered additional information, advice, support, and further intervention as needed. All randomised patients were approached for follow up assessment six months after injury by independent clinicians blind to their group. MAIN OUTCOME MEASURES: Validated questionnaires were used to elicit ratings of post-concussion symptoms (the Rivermead postconcussion symptoms questionnaire), and changes in work, relationships, leisure, social, and domestic activities (the Rivermead head injury follow up questionnaire). RESULTS: The two groups were comparable at randomisation. Data was obtained at six months on 226 of 577 "control" patients and 252 of 579 "trial" patients (59% were lost to follow up). There were no significant differences overall between the trial and control groups at follow up, but subgroup analysis of the patients with moderate or severe head injuries (posttraumatic amnesia > or = one hour, or admitted to hospital), showed that those in the early intervention group had significantly fewer difficulties with everyday activities (P = 0.03). CONCLUSIONS: The results from the 41% of patients followed up do not support the recommendation of offering a routine follow up to all patients with head injury, but they do suggest that routine follow up is most likely to be beneficial to patients with moderate or severe head injuries. Some of those with less severe injuries do continue to experience difficulties and need access to services. A further trial is under way to test these conclusions.