OBJECTIVE: To describe the characteristics and outcome of patients who have a cardiac arrest at home compared with elsewhere out of hospital. PATIENTS: Subjects were patients included in the Swedish cardiac arrest registry between 1990 and 1999. The registry covers about 60% of all ambulance organisations in Sweden. METHODS: The study sample comprised patients reached by the ambulance crew and in whom resuscitation was attempted out of hospital. There was no age limit. Crew witnessed cases were excluded. The patients were divided into two groups: cardiac arrest at home and cardiac arrest elsewhere. RESULTS: Among a study population of 24 630 patients the event took place at home in 16 150 (65.5%). Those in whom the arrest took place at home differed from the remainder in that they were older, were more often women, less often had a witnessed cardiac arrest, were less often exposed to bystander cardiopulmonary resuscitation (CPR), were less often found in ventricular fibrillation, and had a longer interval between collapse and call for ambulance, arrival of ambulance, start of CPR, and first defibrillation. Of patients in whom the arrest took place at home, 11.3% were admitted to hospital alive, v 19.4% in the elsewhere group (p < 0.0001); corresponding figures for survival after one month were 1.7% v 6.2% (p < 0.0001). The adjusted odds ratio for survival after one month (at home v not at home; considering age, sex, initial arrhythmia, bystander CPR, aetiology, and whether the arrest was witnessed) was 0.40 (95% confidence interval 0.33 to 0.49; p < 0.0001). CONCLUSIONS: Sixty five per cent of out of hospital cardiac arrests in Sweden occur at home. The patients differed greatly from those with out of hospital cardiac arrests elsewhere, and fewer than 2% were alive after one month. Having an arrest at home was a strong independent predictor of adverse outcome. Further research is needed to identify the reasons for this.