Mechanical reinforcement in nanofilled polymer melts and elastomers was investigated with emphasis on low filler concentrations ranging from zero to 12.5 vol. %. Composite samples were prepared using a poly(vinyl acetate) matrix with each of three different filler types. Two fumed-silica fillers with different surface chemistry were used. One had an untreated (not-treated) surface (NT) containing hydroxyl groups that can form hydrogen bonds with the matrix polymer, and one had a surface treatment (ST) that prevents such bonds from forming. In addition, a poly(tetrafluoroethylene) (PTFE) filler was used in order to provide no appreciable filler—matrix bonding. Dynamic mechanical data were obtained within the linear viscoelastic region using symmetric, simple shear specimens. Data are presented in the form of superposition master curves of storage shear modulus versus reduced frequency. The increase in storage modulus with filler addition is strongly dependent on the frequency of observation. At higher reduced frequencies, that is, the glassy end of the master curve, reinforcement R (defined as the storage modulus of the composite divided by that of the neat polymer, at a given reduced frequency) is low and has values typically predicted by micromechanics theory, ca. 1.5 at 12.5% filler. Conversely, at lower reduced frequencies in the rubbery plateau region, reinforcement is many times larger than what current theory predicts, ca. 50 or more. In addition, the long relaxation times of the matrix polymer are extended by filler addition, for example, by more than 2 decades for the 12.5% NT filler composite. The relaxation time distribution of the matrix polymer depends on the filler type and concentration, and the master curves themselves cannot be superposed. Of major importance, it is found that the reinforcement increases exponentially with a reduced filler volume fraction defined as the actual volume fraction ϕ divided by a scale factor ψ, that is, R=exp(ϕ/ψ). This scale factor, or “scaling volume fraction”, is specific to the filler-matrix composite system. The initial slope of the reinforcement versus filler volume fraction curve at zero filler is given by 1/ψ, which varies from 28.6 for NT filler to 3.57 for PTFE filler. These values should be compared to the 2.5 found in the Einstein and Guth-Gold theories, both of which fail to describe the physics of reinforcement in polymeric melts. The implications of these findings to the root mechanism of reinforcement and the origin of the Payne effect are discussed in terms of a postulated filler-induced conformational restriction model based on the physics of the polymeric matrix in the presence of filler.