Gas-to-particle synthesis under high temperature is one of the most important methods for producing multicomponent nanoparticles. The volume enlargement of particles due to aggregation accompanies the component mixing within particles in a nonreactive system. To tailor nanocomposites, it is essential to gain an insight into the dynamic evolution of compositional distributions. In this paper, the differentially weighted Monte Carlo (DWMC) method for population balance modeling is used to simulate the process of aggregative mixing. On the methodological end, a new shift action is proposed to regulate a limited number of simulation particles to be distributed as homogeneously as possible over high-dimensional and inhomogeneous joint space of multiple components, where some simulation particles in less-populated regions are split into more simulation particles in order to increase sample space for stochastic statistics and then fatigue against statistical noise, at the same time a certain number of simulation particles in densely populated regions are randomly removed from the simulation to reduce computational demands. The DWMC with the new shift action is used to simulate the aggregative mixing process of bicomponent nanoparticles with compositional-independent or -dependent Brownian coagulation kernel in the free-molecular regime. It is found that the compositional distributions satisfy self-preserving formulation as the particle size distribution in monocomponent systems; and the extent of the time evolution of the degree of mixing (the mass-normalized power density of excess component A) corresponds with that of self-preserving distributions. The compositional distributions and the degree of mixing predicted by the DWMC agree well with theoretical models, while the constant-number method (using equally weighted simulation particles) fails in the more advanced stages of aggregative mixing.