Susceptibility of polymeric composites to moisture has been well known for several decades. Most high performance epoxy or bismaleimide (BMI) resins and their fiber-reinforced composites may absorb up to 5% wt. moisture which could lead to 10-30% reduction in various mechanical properties, including flexural strength, stiffness, impact resistance, and interlaminar shear strength (ILSS). In particular, fiber-matrix interface and process-induced defects such as microvoids often act as moisture storage sites, thus increasing maximum intake level. It has been common practice to use a one-dimensional Fickian model with or without corrections to characterize and predict the diffusion of moisture into polymeric composites. However, in several high-performance and mission critical applications, more sophisticated models accounting for the edge effects, anisotropy of absorption, molecular interactions, and interfacial storage are required to fully describe the long- and short-term moisture absorption dynamics. In this article, a comprehensive absorption model that combines the anomalous diffusion hindrance due to molecular bonding and three-dimensional anisotropy will be presented. The hindered diffusion model (HDM) is shown to predict both short-term Fickian and long-term anomalous, non-Fickian absorption behavior often observed in structural composites. The total amount of absorption is shown to be the sum of bound and unbound liquids which are coupled through a differential diffusion and a temporal storage model. The accuracy of the model predictions is discussed by comparing the model predictions with the experimentally measured total mass gain of a number of material systems, including graphite/epoxy laminates and clay/epoxy nanocomposites. It is shown that the anomalous moisture absorption dynamics observed in all laminate sizes and thicknesses can be accurately predicted by the hindered diffusion model.