Introduction: Bystander cardiopulmonary resuscitation in out-of-hospital cardiac arrest is associated with higher survival rates. Even trained health care staff cannot assess breathing well enough to detect cardiac arrest. Recognition of cardiac arrest by lay rescuers might be overlooked in adult basic life support resuscitation guidelines, which explain what to do, but not how to do it. The 2015 Adult Advanced Life Support Resuscitation Guidelines recommend to “look for chest movement.” We hypothesize (1) that instructing lay rescuers to look for chest movement allows detecting breathing (or lack thereof); (2) that showing a person their own recorded gaze overlay during a video debriefing intervention enhances breathing detection at postallocation; and (3) that the more time spent looking at a cardiac arrest victim's chest, the greater the probability of detecting breathing (or lack thereof). Methods: Monocentric, blinded, prospective, 2-arm parallel randomized controlled trial with balanced randomization (1:1). The design entailed a preallocation simulation, an intervention (video debriefing with or without gaze overlay), and a postallocation simulation. A follow-up simulation took place after 6 months. The main outcome measured was success in detecting breathing. Participants were all prospective students of a bachelor's degree program in nursing. Results: All participants performed better at postallocation (success rate at preallocation = 59%, postallocation = 79%, χ2 = 7.22, P 0.01) regardless of viewing their own gaze overlay during video debriefing. We failed to obtain a sufficient number of participants for the follow-up simulation. Instructing lay rescuers to look for chest movement allows them to detect breathing (or lack thereof). Each second spent looking at the thorax increased the odds of successfully detecting breathing by 38%. Mean thorax gaze duration significantly increased by 5.95 seconds (95% confidence interval = 4.71–7.31) from preallocation (3.46 seconds, SD = 4.16) to postallocation (9.41 seconds, SD = 5.98). Laypersons' median diagnosis time was 15.5 seconds (range = 2–63 seconds), similar to another study (13 seconds, range = 5–40 seconds). Conclusions: This is the second study in which the median time to decision exceeded the maximum 10 seconds recommended. International guidelines should consider increasing the time allowed for the “check breathing” step of bystander cardiopulmonary resuscitation procedures.