Nanofluid is a liquid suspended stably with nano size particles, and it attracts many research efforts because it has a potential to be an excellent coolant. The nano particles could be metal or non-metal, and of different geometric shapes. Because of the Brownian motion, the particles could collide one another and form agglomerates with time, and might precipitate. In such a situation, the nanofluid loses its ability of heat transfer enhancement. The goal of this thesis is to study the particle agglomeration and its effect on thermal conductivity of nanofluid, and focus on: (1) the time evolution of the particle agglomerates, (2) the effect of particle agglomeration on thermal conductivity, and (3) the size distribution of particles during agglomeration. For item (1), we employ the fractal model in literature for predicting the time evolution of average particle size, and measure the average diameter and zeta potential of the particles. The fractal model fails to predict the time evolution of particle diameter, but a recent model by Lei et al. based on the microscopic motion of particles does predict correctly the experimental findings. Thus the experimental particle diameters, instead of the diameters predicted using fractal model, were incorporated with the model of Prasher et al. for item (2) for predicting the thermal conductivity. The theoretical results agree nicely with the experiments of TiO2-water nanofluids for volume fraction from 1 – 2%. It was found that particle properly agglomerate in nanofluids, which is helpful for its heat transfer. For item (3), simulations were performed based on the one-dimensional general dynamic equation subject to Brownian coagulation, together with the log-normal distribution assumption for particles, and the Crank-Nicolson scheme. It was found that the initial particle number concentration, the interaction forces between particles, and the ionic strength of the nanofluid are crucial for particle agglomeration. It is hoped that the present study can provide us a better understand of the physics of particle agglomeration in nanofluids, which is helpful for further research and application.