[en] OBJECTIVE: Automated external defibrillators (AEDs) are becoming increasingly available in public places to be used by citizens in case of cardiac arrest. Most AEDs are semi-automatic (SAEDs), but some are fully automatic (FAEDs) and there is ongoing debate and concern that they may lead to inadvertent shocks to rescuers or bystanders because the timing of the shock is not controlled by the rescuer. We therefore compared the behaviour of untrained citizens using an FAED or an SAED in a simulated cardiac arrest scenario. DESIGN AND PARTICIPANTS: One hundred and seventy-six laypeople were randomised to use an FAED or an SAED (Lifepak CR+, Medtronic, Redmond, USA) in a simulated cardiac arrest scenario on a manikin (Ambu, Denmark) where a bystander was touching the victim's upper arm. Each rescuer's performance was recorded on video and analysed afterwards using a modified Cardiff Score. The rescuer or the bystander was considered unsafe if either of them touched the victim during shock delivery. RESULTS: Eleven cases could not be analysed because of technical problems. Fifteen participants violated the protocol making further analysis impossible. Of the remaining 150 participants, 68 used the FAED and 82 used the SAED. The rescuers were safe in 97/150 (65%) cases, without a difference between FAED and SAED. The bystander was safe in 25/68 (37%) cases in the FAED group versus 19/82 (23%) in the SAED group (p=0.07). Combined safety of both rescuer and bystander was observed in 23/68 (34%) cases in the FAED group versus 15/82 (18%) in the SAED group (p=0.03). CONCLUSIONS: Safety was not compromised when untrained lay rescuers used an FAED compared with an SAED. The observation of overall safer behaviour by FAED users in the presence of bystanders may be related to the additional instructions provided by the FAED, and the reduced interaction of the rescuer with the bystander when using the SAED.