Objectives To evaluate the effect of a sound simulation lesson to improve cardiac auscultation skills among junior doctors. Methods This study is based on the design of test comparison before and after educational intervention using a convenient sample. For 50 junior doctors in Japan, diagnostic accuracy before and after a sound simulation lesson for cardiac auscultation skills was compared. There were 15 doctors who experienced cardiology rotation. The lesson used seven abnormal cardiac recordings (third heart sound, double gallop, aortic stenosis, aortic regurgitation, mitral stenosis, mitral regurgitation, and pericardial friction rub). At tests before and after the lesson, the doctors listened to random sound outputs of the same seven recordings, chose diagnostic findings from multiple-choice items, and obtained individual diagnostic accuracy based on the total number of choosing correct findings. Top 10 doctors obtaining the greatest individual accuracy received a commendation. Results Pre-lesson diagnostic accuracy was not different between doctors with cardiology rotation training (total diagnostic accuracy of the group, 27/105 [26%]) and those without cardiology rotation (70/245 [29%]). Compared to pre-lesson, post-lesson total diagnostic accuracy significantly improved with about two-folds (97/350 [28%] vs 170/350 [61%]; McNemar Test, p<0.0001). The improvement was significant for double gallop (5/50 [10%] vs. 15/50 [30%]), mitral stenosis (0/50 [0%] vs. 6/50 [12%]), and pericardial friction rub (1/50 [2%] vs. 35/50 [70%]). Conclusions The use of a simple sound simulation lesson may help junior doctors to learn cardiac auscultation skills. Clinician educators are encouraged to use this strategy in addition to cardiology rotation training.