To evaluate the 3.02 software version of the FloTrac/Vigileo™ system for estimation of cardiac output by uncalibrated arterial pressure waveform analysis, in septic shock. Nineteen consecutive patients in septic shock were studied. FloTrac/Vigileo™ measurements (COfv) were compared with pulmonary artery catheter thermodilution-derived cardiac output (COtd). The mean cardiac output was 7.7 L min-1 and measurements correlated at r = 0.53 (P < 0.001, n = 314). In Bland-Altman plot for repeated measurements, the bias was 1.7 L min-1 and 95 % limits of agreement (LA) were -3.0 to 6.5 L min-1, with a %error of 53 %. The bias of COfv inversely related to systemic vascular resistance (SVR) (r = -0.54, P < 0.001). Above a SVR of 700 dyn s cm-5 (n = 74), bias was 0.3 L min-1 and 95 % LA were -1.6 to 2.2 L min-1 (%error 32 %). Changes between consecutive measurements (n = 295) correlated at 0.67 (P < 0.001), with a bias of 0.1 % (95 % limits of agreement -17.5 to 17.0 %). All changes >10 % in both COtd and COfv (n = 46) were in the same direction. Eighty-five percent of the measurements were within the 30 -330 of the polar axis. COfv with the latest software still underestimates COtd at low SVR in septic shock. The tracking capacities of the 3.02 software are moderate-good when clinically relevant changes are considered.