BACKGROUND Assessing left ventricular (LV) systolic function in a rapid and reliable way can be challenging in the critically ill patient. The purpose of this study was to evaluate the feasibility and reliability of, as well as the association between, commonly used LV systolic parameters, by using serial transthoracic echocardiography (TTE). METHODS Fifty patients with shock and mechanical ventilation were included. TTE examinations were performed daily for a total of 7 days. Methods used to assess LV systolic function were visually estimated, "eyeball" ejection fraction (EBEF), the Simpson single-plane method, mean atrioventricular plane displacement (AVPDm), septal tissue velocity imaging (TDIs), and velocity time integral in the left ventricular outflow tract (VTI). RESULTS EBEF, AVPDm, TDIs, VTI, and the Simpson were obtained in 100%, 100%, 99%, 95% and 93%, respectively, of all possible examinations. The correlations between the Simpson and EBEF showed r values for all 7 days ranging from 0.79 to 0.95 (P < 0.01). the Simpson correlations with the other LV parameters showed substantial variation over time, with the poorest results seen for TDIs and AVPDm. The repeatability was best for VTI (interobserver coefficient of variation (CV) 4.8%, and intraobserver CV, 3.1%), and AVPDm (5.3% and 4.4%, respectively), and worst for the Simpson method (8.2% and 10.6%, respectively). CONCLUSIONS EBEF and AVPDm provided the best, and Simpson, the worst feasibility when assessing LV systolic function in a population of mechanically ventilated, hemodynamically unstable patients. Additionally, the Simpson showed the poorest repeatability. We suggest that EBEF can be used instead of single-plane Simpson when assessing LV ejection fraction in this category of patients. TDIs and AVPDm, as markers of longitudinal function of the LV, are not interchangeable with LV ejection fraction.