Next-generation materials for fast ion conduction have the potential to revolutionize battery technology. Metal–organic frameworks (MOFs) are promising candidates for achieving this goal. Given their structural diversity, the design of efficient MOF-based conductors can be accelerated by a detailed understanding and accurate prediction of ion conductivity. However, the polycrystalline nature of solid-state materials requires consideration of grain boundary effects, which is complicated by challenges in characterizing grain boundary structures and simulating ensemble transport processes. To address this, we have developed an approach for modeling ion transport at grain boundaries and predicting their contribution to conductivity. Mg2+ conduction in the Mg-MOF-74 thin film was studied as a representative system. Using computational techniques and guided by experiments, we investigated the structural details of MOF grain boundary interfaces to determine accessible Mg2+ transport pathways. Computed transport kinetics were input into a simplified MOF nanocrystal model, which combines ion transport in the bulk structure and at grain boundaries. The model predicts Mg2+ conductivity in the MOF-74 film within chemical accuracy (<1 kcal/mol activation energy difference), validating our approach. Physically, Mg2+ conduction in MOF-74 is inhibited by strong Mg2+ binding at grain boundaries, such that only a small fraction of grain boundary alignments allow for fast Mg2+ transport. This results in a 2–3 order-of-magnitude reduction in conductivity, illustrating the critical impact of the grain boundary contribution. Overall, our work provides a computation-aided platform for molecular-level understanding of grain boundary effects and quantitative prediction of ion conductivity. Combined with experimental measurements, it can serve as a synergistic tool for characterizing the grain boundary composition of MOF-based conductors.