A new generation continuous-wave transesophageal Doppler (TED) device for cardiac output (CO) monitoring (Accucom 2, Datascope), which displays aortic blood flow velocity in real time, was evaluated by 140 simultaneous comparisons with thermodilution (TD) in 16 patients early after coronary artery bypass surgery. The aim was to determine whether this technologic advancement improves the accuracy of COTED assessment. Absolute COTED values showed a considerable scatter as compared to COTD [COTD = 1.77 + 0.75-COTD (L/min), r = 0.52]. The bias, i.e., the mean of individual CO differences (COTD - COTED) was −0.37 ± 1.70 L/min (SD). In contrast, relative CO changes (ΔCO, % of preceding value) showed a satisfactory agreement between TED and TD [ΔCOTED = 1.04 + 0.91·ΔCOTD (%), r = 0.84, n = 124], the bias (ΔCOTD - ΔCOTED) being −0.66 ± 9.72%. In 8 of 124 situations (6.45%), however, significant COTED changes opposite in direction to that of significant COTD changes occurred. This frequency was significantly greater (P < 0.01) than the ideal frequency of 0%. The agreement between ΔCOTD and ΔCOTED improved (P < 0.05) when the aortic diameter changes induced by changes in mean arterial pressure were considered [ΔCOTEDMAPC = 1.10 + 0.95·ΔCOTD (%), r = 0.87, n = 124]. Compared with previous results, the reliability of the second generation device to monitor relative CO changes was considerably improved. Provided that the aortic blood flow velocity signal was stable and free from any disturbances, the second generation TED device may be regarded acceptable for CO trend monitoring in sedated, paralyzed patients. Absolute COTED values, however, still showed a high scatter compared to COTD; the second generation TED device thus still is unsuitable to accurately measure absolute CO values.