Abstract:  [Objective] Diffusion governs fluid transport and production behavior in shale oil and gas reservoirs with nano–microscale pore–fracture systems. However, strong nanoscale confinement and multiscale structural complexity make the underlying mechanisms unclear, and existing models lack cross-scale predictive capability. This study aims to clarify diffusion mechanisms and develop a model applicable to multiscale porous media. [Methods] High-temperature and high-pressure diffusion experiments were conducted, combined with fractal modeling and nanoscale confinement analysis. A fractal–confinement diffusion model was established and validated against experimental data. [Results] Nanoscale confinement significantly affects phase behavior and diffusion capacity. (1) Confinement induces phase-diagram contraction and a leftward shift of the critical point, leading to a reduction of effective diffusivity by approximately five orders of magnitude compared with bulk fluids. (2) Porosity and permeability are the primary structural controls. When porosity increases from 5.68% to 10.53%, diffusion coefficients increase by about one order of magnitude. When bedding-fracture permeability increases from 1.69 mD to 404.88 mD, diffusion coefficients rise by 3–4 times, indicating enhanced transport pathways. (3) Pressure and temperature exert different effects: increasing pressure (10–36 MPa) suppresses gas-phase diffusion but promotes liquid-phase diffusion, whereas increasing temperature (80–115 °C) enhances diffusion for both phases. (4) Diffusion capacity in fractures exceeds that in the matrix by 1–2 orders of magnitude, demonstrating the dominant role of fractures in large-scale transport. (5) The proposed model matches experimental results well and captures multiscale diffusion behavior, enabling structural sensitivity analysis and cross-scale prediction. [Conclusions] Diffusion in shale reservoirs is jointly controlled by nanoscale confinement, pore–fracture structure, and thermodynamic conditions. Confinement reduces diffusivity and alters phase behavior, while fractures significantly enhance transport capacity. The proposed model effectively integrates these effects and provides accurate cross-scale predictions. [Significance] This study establishes a practical framework for modeling shale-fluid diffusion and provides theoretical support for quantitative evaluation of fluid transport and optimization of reservoir development.