Objective. For automated external defibrillators (AEDs) to be practical for broad public use, responders must be able to use them safely and effectively. This study's objective was to determine whether untrained laypersons could accurately follow the visual and voice prompt instructions of an AED. Methods. Each of four different AED models (AED1, AED2, AED3, and AED4) was randomly assigned to a different group of 16 untrained volunteers in a simulated cardiac arrest. Four usability indicators were observed: 1) number of volunteers able to apply the pads to the manikin skin, 2) appropriate pad positioning, 3) time from room entry to shock delivery, and 4) safety in terms of touching the patient during shock delivery. Results. Some of the 64 volunteers who participated in the study failed to open the pad packaging or remove the lining, or placed the pads on top of clothing. Fifty-percent of AED2 pads and 44% of AED3 pads were not placed directly on the manikin skin compared with 100% of AED1 and AED4 pads. Adjacent pad displacements that potentially could affect defibrillation efficacy were observed in 6% of AED1, 11% of AED2, 0% of AED3, and 56% of AED4 usages. Time to deliver a shock was within 3.5 minutes for all AEDs, although the median times for AED1 and AED4 were the shortest at 1.6 and 1.7 minutes, respectively. No significant volunteer contact with the manikin occurred during shock delivery. Conclusions. This study demonstrated that the AED user interface significantly influences the ability of untrained caregivers to appropriately place pads and quickly deliver a shock. Avoiding grossly inappropriate pad placement and failure to place AED pads directly on skin may be correctable with improvements in the AED instruction user interface.