Objective To assess (1) agreement between the ultrasonic cardiac output monitor (USCOM) 1A device for measurement of cardiac output in newborn infants and conventional echocardiography (ECHO), (2) repeatability of USCOM measurements and (3) agreement between novice and expert users of the USCOM.

Design A prospective observational study.

Setting The Neonatal Unit at the Royal Children's Hospital, Melbourne, Australia.

Patients 56 term and near-term infants, with no evidence of structural or functional cardiovascular disease, or haemodynamic shunts.

Measurements Agreement between ECHO and USCOM was assessed by paired measurements of ventricular outputs by a single experienced user. Repeatability was assessed using five repeated measurements in 10 infants. Agreement between five novices and one expert user was assessed by paired USCOM measurements over 30 training measurements.

Results Agreement between USCOM and ECHO for left ventricular output (LVO) was (bias, ±limits of agreement, mean % error): 14, ±108 ml/kg/min, 43%, and for right ventricular output (RVO): −59, ±160, ml/kg/min, 57%. Intra-observer repeatability was 6.7% for USCOM LVO and 3.6% for ECHO LVO. After five training measurements, the mean difference between USCOM measures of LVO by novice and expert users was less than 50 ml/kg/min, but with variability.

Conclusions Repeatability of USCOM measures is high in newborn infants. New users can be trained quickly, but with high inter-user variability. Agreement between USCOM and conventional ECHO is broad, and worse for RVO and LVO. Further studies are required to assess the ability of the device to detect clinically significant changes in infant cardiac output.