BACKGROUND: Early recognition of cardiac arrest is essential for increasing the likelihood of successful resuscitation. However, many factors could obstruct the recognition of cardiac arrest and delay the delivery of cardiopulmonary resuscitation and automated external defibrillator use. We have developed a new system using infrared light to recognize cardiac arrests during emergency. The aim of this study was to evaluate whether cardiac arrests could be appropriately diagnosed by this system in clinical practice. METHODS: During the initial treatment patients 18 years old and older with unconscious level of 300 on Japan Coma Scale were prospectively registered from May 1st 2016 through May 31st 2017 (University Hospital Medical Information Network-Clinical Trials Registry 000022137). The settings for this study were two critical care medical centers in Osaka Prefecture and two suburban emergency medical services in Chiba Prefecture and Osaka Prefecture in Japan. We evaluated each patient, using the diagnosis of cardiac arrest by relevant physicians or emergency medical services personnel as the "gold standard". Finally, the sensitivity and specificity of the system in understanding whether the patient has cardiac arrest were assessed. RESULTS: Out of 207 unconscious patients, 163 patients were diagnosed as suffering from cardiac arrest and 44 patients were identified as experiencing pulsating cardiac rhythm. The developed system for diagnosing cardiac arrest when used within 10 s from the activation of the system had a sensitivity of 100% and a specificity of 55.2%. Additionally, the system had a sensitivity of 100% and a specificity of 63.6% for diagnosing cardiac arrest when used within 20 s from activation. CONCLUSIONS: The newly developed system has 100% sensitivity in detecting cardiac arrests within 10 s from activation of the system in emergency settings. This developed system could help bystanders to promptly initiate resuscitation.