Different empirical models of soil resistance and soil evaporation efficiency for the soil-atmosphere water exchange have been proposed by many investigators and used in several land surface models. However, ad hoc modifications have to be used when these empirical models are extended to simulate bi-directional soil-atmosphere exchanges of volatile organic compounds. We developed a physically-based top boundary condition that consistently accounts for differences in solubility and diffusivity of volatile chemical species during transport and reaction within the soil and exchange with the atmosphere. We applied this top boundary condition to the bare-soil evaporation problem and found that several widely used empirical soil resistance and soil evaporation efficiency models are flawed in some aspects. We also demonstrated that current field measurements are far from sufficiently accurate to derive accurate empirical formulations of soil resistance. The discrepancy between measurement-derived soil resistance and the theoretical exact soil resistance is severely impacted even when the measurement uncertainty is ~5%.