OBJECTIVE To determine the effect of 5-mL injectate on cardiac output measurements in critically ill patients with low ventricular ejection fraction (< 35%). METHODS Thermodilution cardiac output measurements obtained with three 5-mL and three 10-mL (randomly ordered) iced injectates in 50 patients with low ejection fraction were averaged if the measurements were within 10% of the median. If the 3 measurements were not within those limits, additional measurements were obtained. RESULTS Cardiac output measured with the 5-mL injectate (mean, 4.63 L/min) and cardiac output measured with the 10-mL injectate (mean, 4.52 L/min) were not significantly different (P = .64). Lower and upper limits of agreement were -1.7 L/min to +1.6 L/min. The bias (mean difference between 10- and 5-mL measurements) of all measurements was -0.09, and the precision was 1.43 L/min, with a 95% confidence limit (mean difference +/- 2 SD) of -1.7 to +1.6 L/min. An additional measurement was necessary in 77% of patients in the 5-mL group but in only 48% of the 10-mL group (P = .006). CONCLUSIONS Cardiac outputs measured with 5- and 10-mL injectates do not differ significantly. The greater variability of measurements obtained with a 5-mL injectate suggests that more measurements, and thus more time, are needed to measure cardiac output accurately. Clinicians must weigh the benefit of minimizing fluid volume used against the potential decreased reliability of cardiac output measurements.