We investigate by kinetic analysis the aggregation mechanism of two monoclonal antibodies belonging to the IgG1 and IgG2 subclass under thermal stress. For each IgG, we apply a combination of size exclusion chromatography and light scattering techniques to resolve the time evolution of the monomer, dimer, and trimer concentrations, as well as the average molecular weight and the average hydrodynamic radius of the aggregate distribution. By combining the detailed experimental characterization with a theoretical kinetic model based on population balance equations, we extract relevant information on the contribution of the individual elementary steps on the global aggregation process. The analysis shows that the two molecules follow different aggregation pathways under the same operating conditions. In particular, while the monomer depletion of the IgG1 is found to be rate-limited by monomeric conformational changes, bimolecular collision is identified as the rate-limiting step in the IgG2 aggregation process. The measurement of the microscopic rate constants by kinetic analysis allows the quantification of the protein-protein interaction potentials expressed in terms of the Fuchs stability ratio (W). It is found that the antibody solutions exhibit large W values, which are several orders of magnitude larger than the values computed in the frame of the DLVO theory. This indicates that, besides net electrostatic repulsion, additional effects delay the aggregation kinetics of the antibody solutions with respect to diffusion-limited conditions. These effects likely include the limited efficiency of the collision events due to the presence of a limited number of specific aggregation-prone patches on the heterogeneous protein surface, and the contribution of additional repulsive non-DLVO forces to the protein-protein interaction potential, such as hydration forces.