By employing an idealized model of a polymer network and filler, we have investigated the stress-strain behavior by tuning the filler loading and polymer–filler interaction in a broad range. The simulated results indicate that there actually exists an optimal filler volume fraction (between 23% and 32%) for elastomer reinforcement with attractive polymer–filler interaction. To realize this reinforcement, the rubber–filler interaction should be slightly stronger than the rubber–rubber interaction, while excessive chemical couplings are harmful to mechanical properties. Meanwhile, our simulated results qualitatively reproduce the experimental data of Bokobza. By introducing enough chemical coupling between the rubber and the filler, an upturn in the modulus at large deformation is observed in the Mooney-Rivlin plot, attributed to the limited chain extensibility at large deformation. Particularly, the filler dispersion state in the polymer networks is also characterized in detail. It is the first demonstration via simulation that the reinforcement mechanism stems from the nanoparticle-induced chain alignment and orientation, as well as the limited extensibility of chain bridges formed between neighboring nanoparticles at large deformation. The former is influenced by the filler amount, filler size and filler–rubber interaction, and the latter becomes more obvious by strengthening the physical and chemical interactions between the rubber and the filler. Remarkably, the reason for no obvious reinforcing effect in filled glassy or semi-crystalline matrices is also demonstrated. It is expected that this preliminary study of nanoparticle-induced mechanical reinforcement will provide a solid basis for further insightful investigation of polymer reinforcement.