Soil vapour extraction (SVE) is a widely accepted and cost-effective technique used to remediate unsaturated soils contaminated with volatile organic compounds (VOCs). In order to improve SVE design, it is necessary to develop a comprehensive mathematical model that incorporates multiphase flow and multicomponent transport with nonequilibrium mass transfer. The model must include key controlling parameters such as relative permeability, dispersion coefficients, phase densities and interphase mass transfer. Research has been completed on comprehensive three-dimensional SVE models entitled 3D-SVE-L/F. Use of these models allows quantitative evaluation of the SVE tailing effect, a current obstacle for SVE technology. The numerical solutions of the 3D-SVE-L/F models are obtained using FEMLAB, a commercial multi-physics modeling software developed by COMSOL Inc.; 3D-SVE-L/F have been calibrated against known data from lab-scale and field-scale SVE operations. The numerical simulation study indicates that 3D-SVE-L/F models can simulate SVE tailing effects. Accordingly, the pressure field and the distribution of the concentration of contaminant in the soil gas phase as well as the saturation reasonably are predicted. The completed multivariable sensitivity analysis of the calibrated 3D-SVE-L/F models under a 95% confidence interval manifests that empirical mass transfer parameters consisting of the NAPL to vapour mass transfer coefficient are the most sensitive, followed by air-phase permeability. Dispersivity is the least sensitive. Comparison of the mass transfer coefficients between lab and field has shown that the field conditions are more resistant to mass transfer, consistent with high water content, more complex soil properties, and site heterogeneity. A challenge facing SVE designs is estimating the length of SVE treatment time using a 3D-SVE model; a concept referring to a critical time index (CTI) was developed to predict the closure time for stopping an SVE operation. Applying CTI to an SVE operation may save operational time and cost. Additionally, the 3D model developed in this study can be used to assist in field-scale SVE design.