Survival from out-of-hospital cardiac arrest in cities with populations of more than 1 million has not been studied adequately. This study was undertaken to determine the overall survival rate for Chicago and the effect of previously reported variables on survival, and to compare the observed survival rates with those previously reported. Consecutive prehospital arrest patients were studied prospectively during 1987. The study area was the city of Chicago, which has more than 3 million inhabitants in 228 square miles. The emergency medical services system, with 55 around-the-clock ambulances and 550 paramedics, is single-tiered and responds to more than 200,000 emergencies per year. We studied 3,221 victims of out-of-hospital cardiac arrest on whom paramedics attempted resuscitation. Ninety-one percent of patients were pronounced dead in emergency departments, 7% died in hospitals, and 2% survived to hospital discharge. Survival was significantly greater with bystander-witnessed arrest, bystander-initiated CPR, paramedic-witnessed arrest, initial rhythm of ventricular fibrillation, and shorter treatment intervals. The overall survival rates were significantly lower than those reported in most previous studies, all based on smaller communities; they were consistent with the rates reported in the one comparable study of a large city. The single factor that most likely contributed to the poor overall survival was the relatively long interval between collapse and defibrillation. Logistical, demographic, and other special characteristics of large cities may have affected the rates. To improve treatment of cardiac arrest in large cities and maximize the use of community resources, we recommend further study of comparable metropolitan areas using standardized terms and methodology. Detailed analysis of each component of the emergency medical services systems will aid in making improvements to maximize survival of out-of-hospital cardiac arrest.