Atrial fibrillation (AF) is the leading cause of cardiogenic cerebral embolism (CE), which is the most severe course of ischemic stroke. AF detection is important for preventing the onset of CE, but it is not always easy to diagnose AF before the onset of CE because almost half of the patients with AF are asymptomatic. Here, we report the development of an extremely simple yet highly accurate AF screening system that can measure pulse simply by placing the palm on the detector. Videoplethysmography (VPG) was used in the pulse-measuring device. This system consists of a Windows personal computer (PC) connected to a web camera, and the PC is equipped with a non-contact vital sensing software “Rhythmiru” and a pulse wave analysis program. Pulse wave information was acquired by analyzing video images of the palm. The pulse data were evaluated in two stages using power spectrum analysis and coefficient of variation analysis, based on which AF was determined. The measurement time was less than 1 min, which is satisfactory for practical use. Electrocardiography was performed simultaneously with pulse measurement using this device to evaluate the accuracy of the system. Among 128 patients, 116 were analyzed after excluding duplicate patients and those who had technical problems. The 116 patients included 28 with AF and 88 without AF. The system had 96.4% sensitivity and 97.7% specificity for AF detection. Our method of assessing pulse variability using the palm surface based on VPG information was feasible and highly accurate. AF detection efforts can lead to early detection of asymptomatic AF, resulting in a reduction in CE.