The ability of the time-dependent diffusion-double-reptation (TDD-DR) theory to predict the molecular structure and dynamics of polymer nanocomposites is investigated for poly(butylene succinate) blended with fumed silica particles with contrasting surface treatments (unmodified and modified with silanes). Structural and dynamic parameters such as confined polymer fraction (I center dot (s)) and relaxation time are extracted from fitting the experimental curves for relaxation modulus G(t) by the TDD-DR model with fluctuation effects included. A good fit of experimental data over seven time decades is obtained after modification of the TDD-DR model to account for Rouse relaxation on the short time scale. The fraction of confined polymer extracted from model fitting is in quantitative agreement with the value obtained from the specific reversing heat capacity for poly(butylene succinate) (PBS)/fumed silica nanocomposites. Based on parameters deduced from rheological data, we study the influence of surface functionality on the microstructure of polymer matrix. We conclude that increasing the polymer-particle compatibility through introduction of a hydrophobic functionality on the surface of the particles results in increased amount of confined PBS chains and strong immobilization of the PBS molecules. These interface effects are discussed for the first time in terms of TDD-DR model that takes into account the dynamics of bound polymer chains, allowing prediction of the universal nature of the confinement effect and its role in polymer nanocomposite processing and bulk physical properties.