A 3D Computational fluid dynamics (CFD) model of the fumigation process in the Hal Ross Flour Mill of Kansas State University, Manhattan, Kansas, USA, was formulated for prediction of the gas leakage rate to approximate the gas Half-loss time (HLT) during fumigation with Methyl bromide (MB) and Sulfuryl fluoride (SF). The model consisted of external and internal flow domains. The external domain was used to predict stagnation pressures generated by wind impinging on the mill’s walls. The internal domain was used to predict fumigant leakage rates in terms of HLT. Cracks on the mill’s walls represented the effective leakage areas on the internal flow domain. This modeling approach had been used by the authors (Chayaprasert and Maier) in a previous study, but it was simplified and improved in the present study. The primary simplification in the modeling approach was exclusion of the flour mill’s interior details (e.g., milling equipment), reducing the model formulation and simulation computing times. In the previous study, the gas-tightness of the internal flow domain was identified by varying the flow resistance coefficient of the effective leakage areas until the model yielded a HLT value that was close to the one observed from the experimental fumigant concentration data. In the present study, the domain gas-tightness was verified by building pressurization tests. The model was validated using data from one MB and one SF fumigation experiments. The HLTs provided by simulated fumigations were in good agreement with those determined from the experiments. The result of the present study provides further validation to the modeling approach and emphasizes the importance of building pressurization test for accurate HLT prediction. Keywords : Structural fumigation, Half-loss time, Pilot flour mill, Computational Fluid Dynamics (CFD), Building pressurization test