While particle aggregates play a central role in recent models for nanofluid thermal conductivity, the effect of particle diffusion in a temperature field on the aggregation and transport has yet to be studied in depth. The present work separates the effects of particle aggregation and diffusion using parallel plate experiments, infrared microscopy, Monte Carlo simulations, and rate equations for particle and heat transport. The predicted thermal conductivity and viscosity enhancements are compared to determine the favorability of aggregating nanofluids. Experimental data show non-uniform temporal increases in thermal conductivity and are well described through simulation of the combination of particle aggregation and diffusion. The simulation shows concentration distributions due to thermal diffusion causing variations in aggregation, thermal conductivity and viscosity. The aggregation produces an unfavorable nanofluid. An optimum nanoparticle diameter is calculated to minimize settling, thermal diffusion and aggregation. (C) 2010 Elsevier Ltd. All rights reserved.