The convective heat transfer, pressure drop and required pumping power for the turbulent flow of Al2O3 water, TiO2-water and CuO-water nanofluids in a heated, horizontal tube with a constant heat flux are investigated experimentally. Results show that presenting nanofluid performance by the popular approach of plotting Nusselt number versus Reynolds number is misleading and can create the impression that nanofluids enhance heat transfer efficiency. This approach is shown to be problematic since both Nusselt number and Reynolds number are functions of nanofluid concentration. When results are presented in terms of actual heat transfer coefficient or tube temperature versus flow rate or pressure drop, adding nanoparticles to the water is shown to degrade heat transfer for all the nanofluids and under all conditions considered. Replacing water with nanofluid at the same flow rate reduces the convective heat transfer rate by reducing the operating Reynolds number of the system. Achieving a target temperature under a given heat load is shown to require significantly higher flow rates and pumping power when using nanofluids compared to water, and hence none of the nanofluids are found to offer any practical benefits. (C) 2018 Elsevier Ltd. All rights reserved.