Improving low-efficiency artificial forests represents a popular forest hydrological issue, and exploring the optimal stand density (OSD) of low-efficiency artificial forests is an effective method for improving the soil conditions of forestland to prevent the deterioration of ecological function. Water and nutrients were the main limiting factors for vegetation growth. However, relatively few studies addressed the optimization of stand density based on these two factors at the same time. In this study, a total of 176 standard plots (20 × 20 m2) with six stand densities (~500, ~1000, ~1500, ~2000, ~2500 and ~3000 plants·hm−2) were established to investigate the water resources (soil moisture content (SMC), soil evaporation rate (SER), and vegetation transpiration rate (VTR)) and soil nutrient resources (total nitrogen (TN), total phosphorus (TP), total potassium (TK), calcium (CaCO3), organic matter content (OMC)) in low-efficiency Robinia pseudoacacia forestland on the Loess Plateau in western Shanxi, China from June to September each year from 2017 to 2018. The relationships between stand density and water and nutrient resources were analysed with the response surface method (RSM). The RSM results indicated that the OSD averaged 1594 plants·hm−2 and ranged from 940 to 2386 plants·hm−2. The percentage of standard plots with an unreasonable stand density was 35.29%, and 65% of these plots had a value that was higher than the maximum in the range while 35% had a value that was lower than the minimum. These results indicate that the current stand density should be manipulated to fall within the identified OSD range to ensure the normal functions of soil and water conservation in R. pseudoacacia forests. The results of this study serve as a guide for optimizing the stand density of low-efficiency R. pseudoacacia forests in China.