Objective To evaluate the accuracy of a new cardiac output monitor (FloTrac/Vigileo), originally designed for humans, in dogs. This pulse contour cardiac output monitoring system cannot be calibrated and measures cardiac output (Q̇t) from a standard arterial catheter. Study design Prospective experimental trial. Animals Eight adult Beagle dogs weighing 13.1 (9.8-17.1) kg [median (range)]. Methods Anaesthesia in the dogs was maintained using isoflurane. A pulmonary artery catheter and a metatarsal arterial catheter (22 gauge) were placed. Cardiac output was measured simultaneously 331 times by thermodilution and FloTrac technique. A broad spectrum of Q̇t measurements was achieved through alterations of isoflurane concentration, administration of propofol boluses and dobutamine infusions. Agreement between the methods was quantified with Bland Altman analysis and disagreement was assessed with linear mixed models. Results Median (10th and 90th percentile) cardiac output as measured with thermodilution was 2.54 (1.47 and 5.15) L minute(-1) and as measured with FloTrac 8.6 (3.9 and 17.3) L minute(-1) . FloTrac measurements were consistently higher with a mean bias of 7 L minute(-1) and limits of agreement of -3.15 to 17.17 L minute(-1) . Difference between the methods was most pronounced in high Q̇t measurements. Linear mixed models showed an estimated difference between the two methods of 8.05 (standard error 1.18) L minute(-1) and a significant interaction between mean arterial pressure and method. Standard deviation (4.45 higher) with the FloTrac method compared to thermodilution was increased. Conclusion Compared to thermodilution measurements, the FloTrac system was influenced to a higher degree by arterial blood pressure, resulting in consistent overestimation of cardiac output. Clinical Relevance The FloTrac monitor, whose algorithms were developed based on human data, cannot be used as an alternative for thermodilution in dogs.