Abstract: To improve the classification and recovery efficiency of household waste, this study, based on aerodynamic principles, focuses on typical recyclable materials such as iron, aluminum, plastic, and paper, utilizing density differences to achieve simultaneous separation of multiple waste components. This study combines Computational Fluid Dynamics (CFD) numerical simulations and experiments to systematically investigate the motion patterns and separation characteristics of four types of recyclable waste during air classification. Using particle size, inlet air velocity, and inlet angle as design variables, and classification collection efficiency as the evaluation metric, an orthogonal experimental design and range analysis method were employed to obtain an optimized parameter combination. The results show that under the conditions of a particle size of 5 mm, inlet air velocity of 36 m/s, and inlet angle of 14°, the theoretical optimal classification collection efficiency reaches 95.875%. The particle trajectories recorded by a high-speed camera generally align with the simulation results. The actual classification collection efficiencies obtained from experiments for spherical and flake-shaped materials were 93.300% and 91.250%, respectively, confirming the effectiveness of the density-based air classification method. The findings of this study can provide a basis and reference for optimizing the process parameters of air classification equipment and improving the recycling rates of municipal and rural household waste.