All trauma cases flown over a 3.5 year period by the Metropolitan Helicopter Ambulance (MHA) from the accident scene to the Alfred Hospital were analysed. The MHA carries paramedics trained in advanced life support and is not under direct medical control. There were 254 patients (226 males, 28 females, mean age 34 years) of whom 242 had sustained blunt trauma. The mean distance from the accident scene to hospital was 28 nautical miles. The mean time from dispatch of the MHA to arrival at the Alfred was 82 min. The mean ground time at the scene was 32 min. Major trauma (an injury severity score (ISS) of 15 or more) was present in 62% of patients, and the mean ISS was 22.4. The major treatments at the accident scene by the paramedics were insertion of an intravenous (i.v.) cannula (242 cases), application of splints (197 cases), endotracheal intubation (35 patients) and needle thoracostomy to exclude tension pneumothorax (18 cases). There were 25 patients with a Glasgow Coma Score (GCS) less than 8 who were not intubated at the scene. Review of paramedic management identified four cases where prehospital care could have been improved but it is unlikely the final outcome would have changed: delay in transport (1 case), inadequate i.v. fluid resuscitation (2 cases) and delay in intubation (1 case). There was 1 case of undiagnosed tension pneumothorax that contributed to the patient's death and 1 case of non‐intubation where the outcome may have been altered. Overall there were 38 deaths (14% mortalitjl), which was not significantly different from the predicted mortality of 17%. Paramedics who followed specific protocols, but who were unable to communicate directly with the Alfred Hospital, performed well in the selection of injured patients most likely to benefit from helicopter transport. The prehospital treatment of major trauma patients was good, except for the high (42%) non‐intubation rate in patients with a GCS less than 8.