Laboratory experiments of Miyazaki (1976) were numerically analyzed to study movement of water vapor from humid warm air into initially dry cold soil. A 10-cm high sand column had temperatures of 37 °C at the surface and 20 °C at the bottom and the initial volumetric water content of 0.0045 cm3/cm3. Vapor entered into the soil column at its warmer end, moved downward due to the temperature gradient, and condensated at its colder end. Vapor condensation increased the water content near the bottom of the column where the relative humidity was close to unity. Consequently, liquid water moved upward from the bottom due to the pressure head (moisture) gradient. Furthermore, liquid water evaporated in the middle of the column where the relative humidity started decreasing (< 99.98 %). Water vapor, resulting from evaporation in the column, moved also downward due to the temperature gradient, resulting into water-vapor circulation in the soil column with accompanying phase change. The experiment was evaluated using a modified HYDRUS-1D, which implemented the coupled water, vapor, and heat movement model of Philip and de Vries. This model considers four contributions to the total water flux, i.e., liquid and vapor fluxes due to pressure and temperature gradients. Simulated water content profiles agreed reasonably well with observed profiles. Calculated water contents were slightly smaller near the bottom and greater in the middle of the column. Lowering of the unsaturated hydraulic conductivity decreased the upward liquid flux, improving the agreement between modeled and measured water content profiles. Larger values of the enhancement factor increased the downward vapor flux and the condensation rate. The upward liquid water flux consequently also increased because of the increased pressure head gradient. These two opposing effects compensated each other, making the water content calculations less sensitive to the enhancement factor.