Background This study aimed to determine whether a focused 2-day cardiac ultrasound training course could enable physicians to obtain and interpret focused cardiac ultrasound (FCU) images from critically ill patients. Methods We retrospectively reviewed the FCU images submitted by the physicians who attended a 2-day FCU training courses. Three experienced trainers reviewed the images separately. They determined whether the images were assessable and scored the images on an 8-point scale. They also decided whether the physicians provided correct responses for visual estimations of the left ventricular ejection fraction (LVEF) and right ventricle (RV) dilatation and septal motion. Results Among the 327 physicians, 291 obtained images that were considered assessable (89%). The scores for parasternal short-axis view were lower than those obtained for other transthoracic echocardiographic views, p < 0.001. More physicians provided incorrect appraisals of LVEF than of RV dilatation and septal motion (19.9% vs. 3.1%, p < 0.001). The percentages of incorrect answers by LVEF category were as follows: 34.8% on images of LVEF < 30, 24.7% on images of LVEF 30-54, and 16.4% on images of LVEF >= 55%, p < 0.001. A logistic regression analysis showed that patients with abnormal LVEF were associated with physicians' incorrect assessment of LVEF, with an odds ratio of 1.923 (95% confidence interval (CI):1.071-3.456, p = 0.029). Conclusions A large proportion of physicians could obtain and interpret FCU images from critically ill patients after a 2-day training course. However, they still scored low on the parasternal short-axis view and were more likely to make an incorrect assessment of LVEF in patients with abnormal left ventricular systolic function.