Porous materials are extensively used as catalysts, sorbents, and matrices for the controlled release or uptake of, e.g., pharmaceutics, all of which are either influenced by or rely on diffusion. Diffusion plays a role at two levels. A first level is that of the smallest pores, the nanopores; a second that of the broader mesoand macropores, i.e., the Autobahn network of transport channels that bring molecules in and out of the porous material. In the nanopores, the first level, diffusing molecules constantly feel the walls, and the geometry and chemical composition of the wall considerably influence their motion. Modelling diffusion in nanopores remains an important research topic, since the behaviour is so versatile, while overall continuum descriptions of diffusion in nanoporous materials are lacking. Because the interpretation of measurements is not easy either, molecular simulation is the ideal tool to study diffusion in nanopores. We have studied diffusion in a variety of nanoporous materials: Knudsen diffusion of gases in mesopores, activated diffusion in zeolites, water and solutes in protein crystals and in the ion and water channels that traverse cell membranes. Interestingly, unifyingmodels describing the overall behavior useful to chemical engineers could be derived in each case. At the second level, the network of transport channels can be optimized to minimize diffusion limitations, say, in heterogeneous catalysis. Optimal control theory, multi-scale simulation methods, continuum and network theory can be applied to find the network that, e.g., maximizes chemical yield. Progress in nano-structuring porous materials should allow synthesizing materials with the optimized pore channel network, and a desired structure at the nanoscale. One example is the synthesis of zeolite composites. A fascinating example is the prospect to imitate the remarkably high-flux, highly selective protein channels found in cell membranes: molecular simulations could guide us in selecting which chemical and geometrical structure leads to the desired separation behavior.