Thirty-one patients who were in coma or persistent vegetative state two weeks after sustaining a severe head injury were entered into a coma arousal programme. The coma arousal protocol called for a sequence of vigorous multisensory stimulations to be applied to the patient by a relative for up to eight hours a day for seven days a week. An independent study team monitored two patient outcomes, the time taken to obey a simple command on two consecutive occasions 24 hours apart and patients' score on the Glasgow Outcome Scale 10-12 months post-injury. Outcomes were compared with an historical reference group chosen from the literature, consisting of 135 similarly classified patients. Differences between the pilot study and the reference group patients on initial characteristics suggested that the pilot study patients might have the more favourable outcomes, independent of treatment effect. The sample size was sufficient to detect a 40% improvement in recovery rate. No significant improvements were noted in either the time to obey a simple command (p greater than 0.2) or in the Glasgow Outcome Scale (p greater than 0.25), although the observed difference in the latter group was 11% in favour of the pilot study patients. This study was unable to find any evidence that coma arousal, for all its arduous patient contact, had a markedly better outcome compared with conventional treatment.