To understand how soil water evaporates from soil columns different in layered structure and soil columns homogeneous in soil texture, five soil columns were prepared by filling soil into Plexiglas cylinders, 90 cm in height, 23.5 cm in inner diameter and 0.9 cm in thickness of the wall. Three of them were packed with sandy soil and sandy loess, layer by layer alternatively but different in thickness of the layers (11.25, 22.5 and 45 cm) and two with sandy soil and sandy loess separately and homogeneously. The soil columns were irrigated till they were saturated and then let drain with a 2 cm water layer at their bottoms for 30 days and then they were ready for the evaporation experiment. The columns were sealed at the bottom to ensure zero flux from the lower end. During the experiment, cumulative evaporation, relative evaporation rate and profile water content were monitored and AE/PE ratios (actual evaporation rate / potential evaporation rate) were calculated and meanwhile, evaporation processes from the five soil columns were simulated and analyzed using the HYDRUS-1D model and optimized soil hydraulic parameters based the drainage processes of the two homogeneous soil columns. The soil hydraulic parameters were obtained through optimization of the calculation of profile water contents during the drainage processes in the two homogeneous soil columns. Then the optimized hydraulic parameters were used to simulate drainage processes in the three layered soil columns with much better effect that lowered the relative error by 1% to 9%. Besides, the optimized hydraulic parameters based on variation of profile water content in the two homogeneous soil columns during the drainage processes were used to simulate evaporation processes of the soil columns. Actual evaporation from the columns was determined by weighing the columns on D 0, 2, 4, 9, 14, 22, 30, 34, 40, 45, 49, 53 and 56. Moreover, 12 cm long TDR probes were placed in the columns, one every 10 cm to measure profile water content on D 0, 8, 14, 22, 30, 40 and 56. Analysis of the measurements and simulations show that in the homogeneous column of sandy loess the first phase of evaporation lasted 34 days, and as a result, its cumulative evaporation was the highest, reaching 110.8 mm, while in the homogeneous column of sandy soil and the other three layered columns, the first phase evaporation lasted only one day, so their cumulative evaporation was as low as around 6 mm. The layered soil column with sand on the surface was relatively stable in profile water content, indicating that sand mulch could dramatically reduce soil evaporation. In addition, the three layered soil columns did not differ much in soil evaporation, demonstrating that an 11.25 cm-thick overlay of sandy soil is enough to reduce soil evaporation significantly. Although the simulations agreed well with the observations, the simulation of soil evaporation could be further improved in precision by using drainage-process-based optimized hydraulic parameters and the HYDRUS-ID model. Improving the accuracy of the simulation of soil water characteristic curve and soil hydraulic conductivity and taking into account the coupling effect of water, vapor and heat in soil columns is the major approach to reducing simulation deviation. The findings in the experiment may have some important significances in guiding soil water management in arid and semiarid regions.