The present study was designed to assess a new non invasive method for measuring cardiac output. The thoracic electrical bioimpedance method was compared with the reference one, thermodilution. The measurements were made simultaneously with NCCOM3 (bioimpedance) using the freeze data mode, and with a Swan-Ganz catheter and a haemodynamic computer (thermodilution). The study involved 11 spontaneously breathing patients in a steady haemodynamic state. Ten measurements were carried out with both methods for each patient. Statistical analysis of the 110 paired values was carried out by computer. The various statistical tests applied confirmed that there was a highly significant correlation between values for cardiac output obtained by each of these two methods (r = 0.818; p less than 0.005); they also showed a significantly more important dispersion of the measures for each patient with thermodilution. The mean value of the thermodilution "standard deviation" (0.64 l.min-1) was significantly more important (p less than 0.005) than the one with NCCOM3 (0.24 l.min-1). Thoracic electrical bioimpedance appeared a safe method for measuring cardiac output, providing the limits of the method are kept. The objective is not to replace the Swan-Ganz catheter, but to propose an alternative method for measuring cardiac output. This method is very interesting in many circumstances, particularly for intensive care patients: it is a non invasive technique, continual measurement is not time-limited, and its use is very easy.