To validate automated boundary detection measurements of left ventricular volumes, cardiac output, and ejection fraction, we studied 50 patients in the intensive care unit. End-diastolic volume, end-systolic volume, and ejection fraction were calculated by automated boundary detection and compared with two-dimensional echocardiographic images. Automated boundary detection-derived cardiac output was compared with thermodilution measurements and Doppler calculations of flow through the aortic and pulmonic valves. Automated boundary detection agreed well with two-dimensional measurements for end-diastolic volume (r = 0.98), end-systolic volume (r = 0.98), and ejection fraction (r = 0.91). Cardiac output derived from automated boundary detection correlated with two-dimensional echocardiographic measurements (r = 0.84), thermodilution (r = 0.83), aortic valve Doppler (r = 0.75), and pulmonic valve Doppler (r = 0.60). Automated boundary detection measurements of left ventricular volumes, ejection fraction, and derived cardiac output are feasible in patients in intensive care units. This method yields rapid, accurate result compared with thermodilution, two-dimensional images, and Doppler measurements.