A wide range of invasive and non-invasive techniques for monitoring the haemodynamic condition of critically ill patients is now available. A general reluctance on the part of paediatric intensive care specialists to use pulmonary artery thermodilution catheters and the need for constant realignment of hand-held Doppler probes has necessitated the search for a technique which is relatively non-invasive and provides continuous information on the haemodynamic condition of critically ill paediatric patients. We sought to establish if transoesophageal Doppler fulfilled these criteria. Eleven children who had recently undergone cardiac surgery were studied. Median age was 39 months and weight 14.9 kg. Five simultaneous pairs of measurements of cardiac index (CI: thermodilution) and minute distance (MD: transoesophageal Doppler) were made, as a baseline, when each child was haemodynamically stable. Following a fluid challenge, five repeat pairs of measurements were made. The mean percentage changes for CI and MD were 16.4% (range 5.3-44%) and 16.6% (3.4-47.7%), respectively. The average coefficients of variation for measurements of CI and MD were 3.5% and 2.9%, respectively. The mean difference in percentage change between CI and MD was -0.5% (95% confidence interval for the bias -4% to 3%; limits of agreement -10.7 to +9.7%). Our study indicates that transoesophageal Doppler is reproducible, easy to use and provides clinically acceptable information when following changes in CI in haemodynamically stable paediatric patients.