The enthalpic incompatibility of organic polymer matrices and high surface energy inorganic nanoparticles often leads to phase separation in polymer nanocomposites (PNC) precluding the realization of anticipated property enhancements. The grafting of polymer chains to nanoparticles holds promise as a means for controlling dispersion. A single population of polymer chains with tunable graft density (sigma) and molecular weight (N) is observed to have antithetical enthalpic and entropic effects on interface compatibility. We report the use of bimodal polymer brushes with fundamentally decoupled enthalpic and entropic parameters. Bimodal polystyrene brushes were grafted from 15 nm colloidal silica nanoparticles by a previously reported consecutive RAFT (reversible addition-fragmentation chain transfer) polymerization technique. The combination of a high graft density short brush and a low graft density long brush was found to cause improved nanoparticle dispersion in a polystyrene matrix when compared to single populations of long and short brushes of corresponding graft densities and molecular weights. A new quantitative model was developed to understand these results and was found capable of predicting dispersions in grafted nanoparticle composites in the allophobic dewetting regime. The bimodal-brush-graft particles were also found to be remarkably well dispersed in an entropically unfavorable higher molecular weight matrix. This facilitated a study of the role of matrix-brush entanglement on the thermomechanical properties of PNCs, isolated from the effects of particle dispersion. The best enhancements in glassy properties resulted from improved matrix-brush entanglement, attained by lowering the long chain graft density and increasing the long chain to matrix molecular weight ratio.