Thermophoresis is the realization of the averaged Brownian motion of particles in a fluid, which is subject to a steady temperature gradient. At sufficiently long times, the stronger molecular impulses in the hotter fluid region drive particles towards the colder region, where the molecular impulses are weaker. The effect of the molecular impulses on the particles is described by a stochastic Brownian force. When this force is applied to an ensemble of particles the thermophoretic velocity is the average velocity of the ensemble. In this study the motion of an ensemble of 4000 spherical nanoparticles with the material properties of CNT, aluminum, aluminum oxide, copper and gold was simulated in four base liquids-water, ethyl glycol, engine oil and R134a. The ensemble-averaged results generate the thermophoretic velocity of these particles in the base liquids. It was observed that the computational results agree very well with the few experimental data available for liquids. The computational method is general and may be applied to all heterogeneous systems of nanoparticles in liquids. The numerical results yield very useful information on the process of thermophoresis in liquids as well as values of the thermophoretic coefficients in nanofluids. (C) 2014 Elsevier Ltd. All rights reserved.