BACKGROUND: Information from the Swedish Cardiac Arrest Registry was used to investigate: (a) The proportion of patients suffering an out-of-hospital cardiac arrest who were given bystander cardiopulmonary resuscitation (B-CPR). (b) Where and by whom B-CPR was given. (c) The effect of B-CPR on survival. METHOD: a prospective, observational study of cardiac arrests reported to the Swedish Cardiac Arrest Registry. Analyses were based on standardised reports of out-of-hospital cardiac arrests from ambulance organisations in Sweden, serving 60% of the Swedish population. From 1983 to 1995 approximately 15-20% of the population had been trained in CPR. RESULTS: Of 9877 patients, collected between January 1990 and May 1995, B-CPR was attempted in 36%. In 56% of these cases, the bystanders were lay persons and in 25% they were medical personnel. Most of the arrests took place at home (69%) and only 23% of these patients were given B-CPR in contrast to cardiac arrest in other places where 53% were given CPR. Survival to 1 month was significantly higher in all cases that received B-CPR (8.2 vs. 2.5%). The odds ratio for survival to 1 month with B-CPR was in a logistic regression analysis 2.5 (95% CI 1.9-3.1). CONCLUSIONS: In Sweden, the willingness and ability to perform B-CPR appears to be relatively widespread. More than half of B-CPR was performed by laypersons. B-CPR resulted in a two to threefold increase in survival.