Carbon nanotubes (CNTs) have a unique combination of mechanical properties, including extremely high elastic modulus, large strength, and relatively low bending rigidity. The porous CNT materials also have a potential for application as membranes for gas separation, purification, and sensing. The load transfer in networks of pristine CNT, however, is weak and CNT materials exhibit much smaller stiffness and strength than it can be anticipated based on properties of individual CNTs. Weak mechanical properties prevent CNT materials from using in multiple emerging applications. Covalent cross-linking via irradiation by energetic ion or electron beams, as well as chemical functionalization, allow one to enhance stability of CNT materials and improve their elastic and inelastic properties. In this work, the mechanical and gas transport properties of pristine and cross-linked CNT materials are studies theoretically based on a mesoscopic computational model. This model is capable of predicting the mechanical loading of CNT films and aerogels and gas transport in these materials in large-scale simulations involving thousands of nanotubes. In the mesoscopic model, every nanotube is represented by a chain of cylindrical mesoscopic segments. The mesoscopic force field accounts for the stretching, bending, and buckling of nanotubes, van der Waals interaction between CNTs, and presence of cross-links. Discrete cross-links between nanotubes are described by a model based on the Morse potential function for an individual cross-link and a geometrical model of the effective cross-link bond. The molecular dynamics simulations of pulling-out of a CNT from a cross-linked CNT bundle are performed in order to characterize the effect of cross-links on the shear load transfer between CNTs. The obtained results in the form of force-displacement curves and rates of cross-link breaking and reformation are used to parameterize the mesoscopic model of cross-links. The developed atomistic and mesoscopic models as well as novel continuum-level model of cross-linked CNT bundles are first applied to study the shear lag effect during pulling-out of CNTs from long nanotube bundles. The results of pull-out tests obtained with these models agree with each other and reveal a strong shear lag effect, when the effective sheaf force is limited by stretching of individual nanotubes. This effect is found to be a factor that puts limits on improving the shear load transfer by covalent cross-links. The developed mesoscopic model is then applied to study mechanical properties of thin quasi-two-dimensional films and isotropic three-dimensional aerogels composed of single-walled nanotubes. The in silico generated material samples with the entangled network of bundles of nanotubes are prepared by self-assembly of initially dispersed CNTs in dynamic mesoscopic simulations. The equilibrated films and aerogels are subjected to compressive and stretching loading. The simulations show a few distinct modes of compression of CNT films, including the wrinkling mode, and reveal the mechanisms responsible for irreversible material deformation and fracture by means of cross-link breaking and re-arrangement of nanotubes. In addition to quasi-static loading, elastic wave propagation through the CNT films with a size up to a few micrometers is studied in order to calculate the acoustic speed in CNT network materials. The elastic moduli, ultimate strength, and acoustic speed in CNT films are found as functions of the material and cross-link densities, CNT length, and strain rate. The study of gas diffusion through CNT network materials is based on a hybrid model, including the mesoscopic model for CNTs and a molecular dynamics-type approach based on direct solution of equations of motion for individual gas molecules. This hybrid approach is used to study self-diffusion of argon inside CNT networks and simulate gas permeability of CNT materials placed in between low- and high-pressure gas reservoirs. The simulations reveal the strong effect of porous structure and density of the CNT materials on gas self-diffusivity and the material effective permeability. This work is supported by NASA through an Early Stage Innovations grant from NASA’s Space Technology Research Grants Program (project NNX16AD99G) and by NSF (CAREER project CMMI-1554589). Computational support is provided by the Alabama Supercomputer Center and NASA's Advanced Supercomputing (NAS) Division.