Since 1980 an Emergency Medical Service (EMS) system with a two-tier ambulance service has been operating in Göteborg. During this 12-year period, all cardiac arrests outside hospital have been monitored. Cardiopulmonary resuscitation (CPR) training for the general public began in 1985 and, by the end of 1992, 125 000 persons had been trained. The aim of this study was to define the factors associated with an increased chance of survival after cardiac arrest witnessed out-of-hospital and secondary to ventricular fibrillation. The study group comprises all patients with cardiac arrest witnessed outside hospital in Göteborg between 1980 and 1992, in whom CPR was initiated by our EMS and ventricular fibrillation observed at the first ECG recording. In a multivariate analysis of age, sex, time of collapse, interval between collapse and first defibrillation, bystander-initiated CPR, the following factors were associated with an increased chance of being discharged from hospital: (1) Short interval between collapse and first defibrillation (P < 0.001); (2) Bystander-initiated CPR (P < 0.001); and (3) Age (P < 0.05). Among patients with an out-of-hospital cardiac arrest who were found by the EMS personnel to have ventricular fibrillation, the predictors of survival were: interval between collapse and defibrillation, bystander-initiated CPR and age.