Morbidly obese patients are at increased risk of intra-operative haemodynamic instability, which may necessitate intensive monitoring. Non-invasive monitoring is increasingly used to measure cardiac output; however, it is unknown whether the weight-based algorithm utilised in these devices is applicable to patients with morbid obesity. We compared the level of agreement and trending ability of non-invasive cardiac output measurements (Nexfin(®) ) with the gold-standard thermodilution technique in 30 morbidly obese patients undergoing laparoscopic surgery. Bland-Altman analysis revealed a mean (SD) bias of 0.60 (1.62) l.min(-1) (limits of agreement -2.67 to 3.86 l.min(-1) ) and the precision error was 46%. Polar plot analysis resulted in an angular bias of 2.61°, radial limits of agreement of -60.08° to 49.82° and angular concordance rate was 77%. Both agreement and trending were outside the Critchley criteria for the comparison of cardiac output devices with a gold-standard. Nexfin has an unacceptable level of agreement compared with thermodilution for cardiac output measurement in morbidly obese patients.