We aimed to determine whether our results were any better or worse than other published reports and to examine the efficacy of the West Midlands Ambulance Service (WMAS) policy of applying cardiopulmonary resuscitation (CPR) and manual ventilation to all unwitnessed cardiac arrests in preference to immediate defibrillation. All cardiac arrests were studied from October 1994 to September 1996. In all unwitnessed arrests, crews undertook CPR and manually ventilated the lungs via a mask or an endotracheal tube with a bag and valve or a mechanical resuscitator using an FIO2 of 1 or 0.21 for at least 2 min before defibrillation was attempted. There were 3403 diagnosed cardiac arrests but, in these, the diagnosis was not certain. CPR and advanced life support (ALS) were applied in 3380 patients and return of spontaneous circulation (ROSC) was obtained in 554, giving a success rate of 16.4%. A total of 364 patients were accepted into hospital, 90 patients died in A&E but 274 patients were admitted to ICU/CCU. Seventy died within 24 h, 69 died after 24 h and 135 were discharged alive and well without cerebral damage. The final success to discharge rate was 49.27%. Of those discharged, 69 had a circulatory arrest period of more than 4 min but in only 10 was a bystander available to start CPR. The European Resuscitation Council Guidelines recommending immediate defibrillation for unwitnessed arrests are not supported by these results. The apparent lack of cerebral damage and the percentage success suggests that resuscitation considerations should be as brain orientated as they are heart orientated. The elapsed time periods reported challenge several shibboleths.