This paper describes power performance measurements undertaken on a model floating wind turbine installed on a tension-leg platform (TLP) in a wind/wave generator facility. Initially, the surge of the platform was measured under different rotor and wave conditions. The surge behaviour depended considerably on the rotor tip speed ratio and the wave frequency and amplitude. High-frequency data sampling techniques were then used to derive the instantaneous power coefficient and tip speed ratio directly from the measurements, together with the surge velocity of the floating system. The power measurements were compared with those predicted by three independent numerical models, two of which are based on the blade element momentum approach and the third involving a lifting-line free-wake vortex model. The fluctuations of the power coefficient with time predicted by the three models were in close agreement; however, these were all significantly larger than those derived from the rotor shaft torque measurements. This was found to be due to the limitations of the torque measurement technique. Although being accurate in measuring the time-averaged torque, the sensor was incapable of measuring the high-frequency low-amplitude fluctuations in the rotor shaft torque induced by the TLP surge. This was confirmed using an alternative experimental technique involving hot-wire near-wake measurements. The study also investigated the influence of the platform surge motion on the time-averaged power coefficients. Both the measurements and the free-wake vortex model revealed marginal deviations in the time-averaged power coefficients when compared with those obtained for a fixed, non-surging rotor. Copyright © 2014 John Wiley & Sons, Ltd.