This article discusses the application of a computational fluid dynamics (CFD) approach for constructing a numerical methodology based on Reynolds-Averaged Navier-Stokes (RANS) equations in order to accurately determine particle deposition rates against surfaces subjected to a low-velocity turbulent airflow. We began by studying a horizontal duct flow configuration and validating the numerical output for airflow pattern and particle deposition rate predictions against published numerical data. The main outcome of this first part of the study was that when starting from a one-way coupling Lagrangian formulation, accurate prediction of particle deposition rates strictly requires the construction of a sufficiently fine mesh (characterized by a dimensionless "distance to wall" criterion y + lower than 4) and the use of a turbulence model accounting for turbulence anisotropy in combination with a near-wall "two-layer zonal" boundary condition. The CFD methodology set up was then used to predict airflow patterns together with airborne deposition of spherical 1-50 µm particles against all the surfaces in an 87 m 3 ventilated cold room filled with 9 big rectangular parallelepipeds. The results demonstrated close agreement in air velocity predictions and showed particle deposition rates varying with both room wall surface type and orientation and with particle diameter due to the different physical phenomena involved. This paper is the first attempt to predict particle deposition in such a large volume, corresponding to a food-processing environment.