Tritium retention in plasma-facing materials is a critical issue that can significantly impact the long-term and steady-state operation of fusion devices. The experiments conducted in the laboratory device MIES have confirmed that the presence of the tungsten (W) nanostructure (called “fuzz”) leads to a substantial retention of hydrogen isotopes within W fuzz layer. This observation motivates us to conduct dedicated modelling to investigate the influence of W nanostructures on deuterium (D) retention using the three-dimensional kinetic Monte Carlo code SURO-FUZZ. The SURO-FUZZ code offers a great flexibility in generating diverse microscopic structures of the W fuzzy surface through the quartet structure generation set (QSGS) approach, which allows us to explore the effects of the pore size and shape on D retention. In this study, several different W nanostructures generated by QSGS approach are utilized to conduct a comprehensive comparison between MIES experiments and SURO-FUZZ simulations. It is demonstrated that the simulated D retention can be brought into a reasonable agreement with the experimental data. On this basis, predictive estimations of D retention on EAST and ITER have been performed with SURO-FUZZ modelling. The simulation results indicate that the total D retention induced by W fuzz remains well below the administrative limit of 700 g.