Carbon nanotubes (CNTs) are one of the wonders of modern science. Discovered a little over 15 years ago, they have shown the research community an outstanding set of properties. In terms of mechanical properties, they exhibit extremely high young's modulus, which, coupled with a high strain to break, leads to unsurpassed strength to break. CNTs also demonstrate superior thermal conductivity, good electrical capacity and high thermal stability. In light of these properties, CNTs are expected to be introduced into a wide variety of new materials aimed at applications for various fields, such as high-performance composites, biological and chemical sensors, magnetic recording, nanoelectronic devices and flat panel displays. One such promising application is CNT-reinforced composite materials, exhibiting the possibility of outstanding mechanical properties. In practice, however, many reports indicate that nanocomposites are weaker or only slightly stronger than the neat resins. Several factors are believed to be the primary source of this discrepancy, namely poor nanotube dispersion in resin, inadequate alignment of the nanotubes, and weak interfacial bonding between nanotubes and resins. As a result, these have become crucial investigation issues for developing high-performance nanocomposites. In this dissertation, fundamental understanding of the interfacial phenomena between carbon nanotubes and polymer matrices are studied. Both molecular dynamics (MD) simulation, an effective approach to investigate nanoscale behaviors, and experimental investigation, are utilized to achieve this goal. First, we examine the interface formation phenomena between a Single Wall Carbon Nanotube (SWNT) and the resin, prior to curing, in the case of the Epon862 resin system. The MD simulation results outline the validity of some of the current theories, such as molecular migration and reduction of molecular mobility of the resin, while they seem to indicate some other mechanisms are not present in this resin system, such as molecular wrapping around the SWNTs. Second, existing MD simulation models of nanotube pullout are analyzed and modified to examine the energy of certain material systems more correctly, and to characterize interfacial shear strength in SWNT/polymer composites. The interfacial bonding and load transfer behaviors between the different SWNTs' configurations (open end, capped end, functionalized end) and three different matrices (polystyrene, polyethylene and Epon862) were examined using the modified models. The results of the modified models effectively reveal the effects of different tube configurations and resin matrices on the interfacial strength during a simulated pullout. Finally, we use MD simulation to investigate the coefficient of thermal expansion (CTE) of individual SWNTs, SWNT ropes, as well as SWNT nanocomposites. Experiments were also carried out in order to gain further insight in the results. It is found that, while the CTE of individual nanotubes is of low negative value, the CTE of the same tubes within a rope or a nanocomposite can significantly change. We also find that SWNTs can be utilized to tailor the CTE of the Epon862 resin system, depending on the functionalization of the SWNTs prior to their introduction in the resin. Finally, a new twisting vibration mode was revealed in SWNT ropes that should prove critical in further SWNT rope studies utilizing MD simulation.