This study demonstrates the effect of different denominators on the survival rate from out-of-hospital cardiac arrest. We retrospectively analyzed data from a cardiac arrest surveillance system in King County, Washington during the years 1976 to 1988, and calculated survival rates using eight different definitions of denominators. The eight survival rates ranged from 16% to 49% discharge from hospital. The denominator for the lowest survival rate included all cases of cardiac arrest for whom emergency medical services personnel started cardiopulmonary resuscitation. The denominator for the highest survival rate included: all cases of presumed cardiac etiology; first recorded rhythm was ventricular fibrillation; collapse witnessed; cardiopulmonary resuscitation started by bystanders within 4 minutes; and definitive care provided within 8 minutes. The definition of cases included in the denominator can dramatically effect the resultant survival rate. There must be national and international agreement about definitions of denominators for valid cross community comparisons.