Abstract The pulmonary artery catheter (PAC) is considered the gold standard for cardiac index monitoring. Recently new and less invasive methods to assess cardiac performance have been developed. The aim of our study was to assess the reliability of a non-invasive monitor utilizing bioreactance (Starling SV) and a non-calibrated mini-invasive pulse contour device (FloTrac/EV1000, fourth-generation software) compared to bolus thermodilution technique with PAC (TDCO) during off-pump coronary artery bypass surgery (OPCAB). In this prospective study, 579 simultaneous intra- and postoperative cardiac index measurements obtained with Starling SV, FloTrac/EV1000 and TDCO were compared in 20 patients undergoing OPCAB. The agreement of data was investigated by Bland–Altman plots, while trending ability was assessed by four-quadrant plots with error grids. In comparison with TDCO, Starling SV was associated with a bias of 0.13 L min−1 m−2 (95% confidence interval, 95% CI, 0.07 to 0.18), wide limits of agreement (LOA, − 1.23 to 1.51 L min−1 m−2), a percentage error (PE) of 60.7%, and poor trending ability. In comparison with TDCO, FloTrac was associated with a bias of 0.01 L min−1 m−2 (95% CI − 0.05 to 0.06), wide LOA (− 1.27 to 1.29 L min−1 m−2), a PE of 56.8% and poor trending ability. Both Starling SV and fourth-generation FloTrac showed acceptable mean bias but imprecision due to wide LOA and high PE, and poor trending ability. These findings indicate limited reliability in monitoring cardiac index in patients undergoing OPCAB.