OBJECTIVE: To compare the assessment of cardiac output (CO) in children using the noninvasive Ultrasound Cardiac Output Monitor (USCOM) with the invasive pulmonary artery catheter (PAC) thermodilution cardiac output measurement. DESIGN AND SETTING: Prospective observational study in a tertiary center for pediatric cardiology of a university children's hospital. PATIENTS: Twenty-four pediatric patients with congenital heart disease without shunt undergoing cardiac catheterization under general anesthesia. MEASUREMENTS AND RESULTS: CO was measured by USCOM using a suprasternal CO Doppler probe in children undergoing cardiac catheterization. USCOM data were compared to CO simultaneously measured by PAC thermodilution technique. Measurements were repeated three times within 5 min in each patient. A mean percentage error not exceeding 30% was defined as indicating clinical useful reliability of the USCOM. CO values measured by PAC ranged from 1.3 to 5.3 l/min (median 3.6 l/min). Bias and precision were -0.13 and 1.34 l/min, respectively. The mean percentage error of CO measurement by the USCOM compared to PAC thermodilution technique was 36.4% for USCOM. CONCLUSIONS: Our preliminary data demonstrate that cardiac output measurement in children using the USCOM does not reliably represent absolute CO values as compared to PAC thermodilution. Further studies must evaluate the impact of incorporating effective aortic valve diameters on CO measurement using the USCOM.