Objective. The San Francisco Fire Department deployed an automated, load-distributing-band chest compression device (AutoPulse, Revivant Corporation) to evaluate its function in a large urban emergency medical services (EMS) service. A retrospective chart review was undertaken to determine whether the AutoPulse had altered short-term patient outcome, specifically, return of spontaneous circulation (ROSC). Methods. AutoPulse cardiopulmonary resuscitation (A-CPR) was used by paramedic captains responding to adult cardiac arrests with an average ±SD response time of 15 ± 5 minutes. The primary endpoint was patient arrival to an emergency department with measurable spontaneous pulses. The manual CPR comparison group was case-matched for age, gender, initial presenting electrocardiogram rhythm, and the number of doses of Advanced Cardiac Life Support medications as a proxy for treatment time. Matching was performed by an investigator blinded to outcome and treatment group. Results. Sixty-nine AutoPulse uses were matched to 93 manual-CPR-only cases. A-CPR showed improvement in the primary outcome when compared with manual CPR with any presenting rhythm (A-CPR 39%, manual 29%, p = 0.003). When patients were classified by first presenting rhythm, shockable rhythms showed no difference in outcome (A-CPR 44%, manual 50%, p = 0.340). Outcome was improved with A-CPR in initial presenting asystole and approached significance with pulseless electrical activity (PEA)(asystole: A-CPR 37%, manual 22%, p = 0.008; PEA: A-CPR 38%, manual 23%, p = 0.079). Conclusion. The AutoPulse may improve the overall likelihood of sustained ROSC and may particularly benefit patients with nonshockable rhythms. A prospective randomized trial comparing the AutoPulse with manual CPR in the setting of out-of-hospital sudden cardiac arrest is under way.