Abstract: Abstract: The preparation of straw-based agricultural water-retaining agent is one of the effective ways to utilize crop resources. It can solve the problems of low soil nutrient content and low utilization rate of cultivated land in arid and semi-arid areas of Northwest Liaoning. However, the characteristics of straw such as dispersed texture, low bulk density, high cost in storage and transportation restrict its large-scale and value-added utilization. As a result, the biomass densification technique has attracted wide attention. To the best of our knowledge, the physical properties of the densified products are influenced by the mixing conditions of raw materials, pressing parameters and types of binders, and addition of high protein yellow soybean meal powder into soil can improve the quality of crops, ripen the soil, kill or inhibit soil harmful bacteria, and achieve the balance of fertilizer supply, while the white sodium polyacrylate as the binder can improve the stability and absorption of the products. Therefore, these two components were mixed with straw blocks proportionally to make samples with different water content in this study, then the mixture was loaded into a compression die (50 mm inner diameter and 80 mm height) that controlled by the WDW-200 electronic universal testing machine. The clamp of the testing machine held the die head and keeps the plumb state. After maximum compression force and compression speed were set by a computer, the machine started to move downward, and densified block was extruded from the die at a room temperature. The quality, diameter and height of the product were measured after a complete relaxation, and the relaxation density was calculated. To explore an optimal forming process of agricultural water-retaining agent, a quadratic general rotary combination design was adopted, with the maximum compression force, moisture content of mixture, compression speed, mass fraction of soybean meal and mass fraction of sodium polyacrylate were test factors. The softwares of Design-Expert 8.0.6 and Excel 2010 were applied for the variance analysis and response surface analysis of experimental results, with the aim to establish mathematical model between the index and each influencing factor, as well as optimize the combination of forming parameters. Furthermore, in the optimum range of relaxation density, the soil condition and climate conditions in Northwest Liaoning were simulated. The water retention capacity of formed blocks with or without the sandy soil medium were evaluated and compared by water release at intervals in a temperature controlled drying oven, and effect of relaxation density was analyzed as well. Results showed that the relaxation density was significantly affected by following four factors (P < 0.000 1), and the degree of importance were maximum compressive force > moisture content of mixture > mass fraction of sodium polyacrylate > mass fraction of soybean meal. Under the condition of relaxation density of 460-540 kg/m3, the optimized combinations of cold forming parameters were maximum compression force of 16-20.59 kN, moisture content of mixture of 8.84%-12.96%, compression speed of 95.56-155.51 mm/min, mass fraction of soybean meal of 16.08%-24.02%, mass fraction of sodium polyacrylate of 4.91%-7.15%. The error was 1.03% in the test verification, which was reliable in prediction. The water-holding capacity of formed block was closely related to relaxation density, the water release increased with drying time, but decreased with the increase of relaxation density at the same temperature. Because the block with larger relaxation density had greater binding force on water, the water retention performance was better. By using the softwares of Excel 2010, it was calculated that the block-sand overall porosity decreased, water release decreased slowly and tended to balance with the double action of filling sand and straw block, so water release of block-sand and drying time conformed to the logarithmic model; while in the absence of sand, the regression coefficients of linear fitting (R2 > 0.98) were greater than those of logarithmic fitting (R2 > 0.96), so it followed the linear model obviously, the relaxation density was larger, the slope of straight line was smaller, the water retention performance was stronger. In summary, the regression coefficients of two fitting equations were all greater than 0.9, water release in the sandy soil medium was slower and smaller than that without the sandy soil medium, which indicated that water retention performance of block-sand was better, the analysis of water retention was reliable and provided a reference for the application of compound agricultural water retention agent.