Simultaneously measured cardiac output obtained by thermodilution (TD), transcutaneous suprasternal ultrasonic Doppler (DOP), CO2-rebreathing (CR) and the direct Fick method (FI) were compared in eleven healthy subjects in a supine position (SU), a sitting position (SI), and during sitting exercise at a workload of 50 W (EX). The agreements between the techniques, two by two, were expressed as the bias calculated as the averaged differences between the techniques. Precision was expressed as the standard deviation of the bias. The overall agreement (bias +/- precision) between TD, DOP and CR respectively and FI were 2.3 +/- 1.6, -0.1 +/- 1.4, and -0.2 +/- 1.1 l/min. TD overestimated cardiac output consistently in SU, SI and EX. DOP was in-accurate during EX and agreed well with FI in SU and SI. CR agreed closely with FI in SI and EX, but values were underestimated in SU. The overall agreement between DOP and CR, respectively, and TD were 2.5 +/- 2.2 and 2.6 +/- 1.6 l/min. The overall agreement between DOP and CR was 0.1 +/- 1.6 l/min. In conclusion, TD overestimated cardiac output compared to the other techniques and the poor agreement has to be taken into consideration especially in measures of low values. The precision of DOP and CR against FI seems to be within clinically acceptable limits, and these methods may provide interchangeable alternatives to the invasive Fick method.