Polymer nanocomposites offer a wide range of promising applications because of their much enhanced properties arising from the reinforcement of nanoparticles. However, further development of such nanomaterials depends on the fundamental understanding of their hierarchical structures and behaviors which requires multiscale modeling and simulation strategies to provide seamless coupling among various length and time scales. In this review, we first introduce some computational methods that have been applied to polymer nanocomposites, covering from molecular scale (e.g., molecular dynamics, Monte Carlo), microscale (e.g., Brownian dynamics, dissipative particle dynamics, lattice Boltzmann, time-dependent Ginzburg–Landau method, dynamic density functional theory method) to mesoscale and macroscale (e.g., micromechanics, equivalent-continuum and self-similar approaches, finite element method). Then, we discuss in some detail their applications to various aspects of polymer nanocomposites, including the thermodynamics and kinetics of formation, molecular structure and dynamics, morphology, processing behaviors, and mechanical properties. Finally, we address the importance of multiscale simulation strategies in the understanding and predictive capabilities of polymer nanocomposites in which few studies have been reported. The present review aims to summarize the recent advances in the fundamental understanding of polymer nanocomposites reinforced by nanofillers (e.g., spherical nanoparticles, nanotubes, clay platelets) and stimulate further research in this area.