This study was conducted to measure emergency medical services (EMS) response times in sudden out-of-hospital cardiac arrests and relate those times to probability of survival in cardiac arrest victims in Las Vegas casino-hotels from January 1993 to June 1996. Times from 911 activation to casino arrival and casino arrival to arrival at patient's side (time to first defibrillatory shock), as well as survival to hospital discharge, were studied with regression analysis. Sixty patients survived (29.3%). Response times to the hotels for survivors and nonsurvivors were similar (4.8 v 5.6 min, P = .44). However, times from arrival at the casino to arrival at the patient's side (5.0 v 6.88 min, P = .01) and elapsed times from 911 activation until first shock (9.88 v 12.46 min, P = .02) were substantially longer for nonsurvivors. Model fitting disclosed that with a 911-to-shock time of 4 minutes, survival probability was 36%. Odds decreased by 5% each minute, to 19% after 23 minutes. Ventricular fibrillation was the most common initial rhythm (187 cases) and was associated with the shortest times from 911 to shock (10.7 +/- 7.8 min). There was a strong trend to increased survival with ventricular fibrillation. The 911-to-shock times in this study are considerably better than in other published reports for large metropolitan EMS systems, but the time from 911 to shock was nearly 3 minutes longer for nonsurvivors, and even those defibrillated at 4 minutes had only a 36% chance of survival. New measures, including use of the automatic external difibrillator, to reduce the "vertical" response are urgently needed.