In this article, we utilize finite element modeling to investigate the effect of nanoparticle agglomeration on the glass transition temperature of polymer nanocomposites. The case of an attractive interaction between polymer and nanofiller is considered for which an interphase domain of gradient properties is developed. This model utilizes representative volume elements that are created and analyzed with varying degrees of nanoparticle clustering and length scale of interphase domain. The viscoelastic properties of the composites are studied using a statistical approach to account for variations due to the random nature of the microstructure. Results show that a monotonic increase in nanofiller clustering not only results in the loss of interphase volume but also obstructs the formation of a percolating interphase network in the nanocomposite. The combined impacts lead to a remarkable decrease of T-g enhancement of clustering nanofillers in comparison with a well-dispersed configuration. Our simulation results provide qualitative support for experimental observations that clustering observed at high nanofiller concentrations negatively impacts the effects of the nanofiller on overall properties. (C) 2011 Wiley Periodicals, Inc. J Polym Sci Part B: Polym Phys 49: 740-748, 2011