This thesis presents a fundamental study of the sintering of supported nanoparticles in relation to diesel oxidation catalysts. The sintering of supported nanoparticles is an important challenge in relation to this catalyst, as well as many other catalyst systems, and a fundamental understanding of the sintering mechanisms of nanoparticles is important for making improvements to their long term catalytic activity. Diesel oxidation catalysts are usually composed of noble metal nanoparticles on a complex three-dimensional high surface area oxide. The complex support structure makes it difficult to directly observe dynamical processes such as particle sintering with the present state of the art microscope techniques, and consequently it is difficult to relate experimental observations and theoretical sintering models. To reduce the complexity, the present study uses planar model catalysts. These are composed of Pt, Pd and bimetallic Pt-Pd nanoparticles supported on a flat and homogeneous Al2O3 or SiO2 surface. By using in situ TEM on the planar model catalysts it was possible to directly monitor the detailed dynamical changes of the individual nanoparticles during exposure to oxidizing conditions at elevated temperatures. The time-resolved TEM images are presented and these offer direct insight into the fundamental dynamics of the sintering process at the nano-scale. For Pt, Pd and bimetallic Pt-Pd nanoparticles it is shown that the sintering process is governed by the Ostwald ripening mechanism in an oxidizing environment. The observations compare well with predictions from mean-field kinetic models for ripening, but deviations are revealed for the timeevolution for the individual nanoparticles. A better description of the individual nanoparticle ripening is obtained by kinetic models that include local correlations between neighbouring nanoparticles in the atom-exchange process. The sintering process was also presented statistically by particle size distributions extracted from the TEM images. The statistical data agreed only partly with the mean-field kinetic models for ripening, but the deviations could be accounted for by including more detailed information into the models, such as an observed size-dependence of the three-dimensional shape of the supported nanoparticles and the local correlations between the nanoparticles.