The reliable characterization of nanoporous carbons is critical to the design and optimization of their numerous applications; however, the vast majority of carbons in industrial use are highly disordered, with complex structures whose understanding has long challenged researchers. The idealized slit pore model represents the most commonly used approximation to a carbon nanopore; nevertheless, it has been only partially successful in predicting adsorption isotherms and fails significantly in predicting transport properties because of its inability to capture structural disorder and its effect on fluid accessibility. Atomistic modeling of the structure has much potential for overcoming this limitation, and among such approaches, hybrid reverse Monte Carlo simulation has emerged as the most attractive. This method reconstructs the structure of a carbon based on the fitting of its experimentally measured pair distribution function and appropriate properties such as porosity while minimizing the energy. The method is shown to be best implemented using a multistage strategy, with the first stage used to attain a deep minimum of the energy and subsequent stages to refine the structure based on the fitting of specific properties. Methods to determine the accessibility of gases based on the atomistic structure are outlined, and it is shown that energy barriers are very sensitive to small differences in the sizes of constrictions and pore entries. The ability to accurately predict macroscopic transport coefficients of adsorbates in nanoporous carbons appears to be the greatest limitation of such models. Overcoming this will require the fitting of properties more sensitive to long-range disorder than the currently used pair distribution and the use of a suitable multiscaling strategy, which is suggested as a future direction for advancing atomistic models. The inclusion of heteroatoms in the structure is also an important area requiring further attention, particularly in the development of computationally efficient force fields incorporating their interactions.