In order to study effectiveness of soil radon mitigation systems in dwellings basements, it is necessary to develop robust numerical models that can estimate time-varying radon concentration in the basements. However, development of such models is a very challenging topic. Indeed, there are three difficult tasks that have to be solved: i°) to characterize soil and foundation materials hydraulic properties, and geometry, length, and position of cracks inside these materials, ii°) to solve discontinuity due to free gas flow in a basement volume embedded within a tortuous porous material, and iii°) to characterize the dwelling occupation mode by the inhabitants. The purpose of this work is twofold: first, to investigate the performance of a meshed-basement model (BM1), compared to classical mass balance models (BM2), in simulating measured fluctuations in time of point-radon concentration in the basement; and second, to study efficiency of soil depressurization systems (SDS) designs in reducing the higher radon concentrations in basements due to higher under-pressurizations.Application of both models to an experimental dwelling showed that BM1 is more efficient than BM2 in that it can better simulate both measured radon concentration and radon-entry-rate. Numerical SDS-scenarios by using BM1 showed that the basement sub-slab depressurization (SSD) alone is a costly solution for radon mitigation. However, scenarios combining SSD with a sump pumping well (SPW) design in the soil adjacent to the wall, showed to be a good alternative because it involves less energy for soil air extraction.